Peer Networks and Educational Migration: Experimental Evidence from India

The intervention connects prospective university students with peer mentors who are educational migrants in Pune and Mumbai, through a program called "Campus Trail".

Peer mentors are recruited among applicants who are current university students in Pune and Mumbai. Majority are educational migrants themselves. Selected mentors attend mandatory training sessions and develop destination-specific curriculum covering practical aspects of educational migration (applications, housing, logistics) and personal dimensions (managing social connections, addressing safety concerns), meant to provide guidance based on mentors' lived experiences.

Mentees are recruited through in-person visits to high schools in Maharashtra where the mentorship program is presented. Interested students apply during these school visits are randomly assigned to either: (1) Control group receiving no intervention, or (2) Treatment group with access to peer networks throughout the academic year.

I match peer pairs by gender and field of study to maximize relevance of the relationship. The intervention requires a total of 3 hours of one-on-one phone conversations between each mentor-mentee pair, spread across as many sessions as they prefer throughout the academic year. Scheduling is flexible and determined by the pairs themselves. Mentors are encouraged to tailor the conversations to individual needs and preferences of the mentees.

2025-09-23

2026-06-30

Primary outcomes includes students' educational migration choices:

1) Realized educational migration choices (index)
2) Hypothetical educational migration choices (index)

Index of realized educational migration choices:
* Indicator equal to one if student migrates out of district
* Indicator equal to one if student is enrolled in college
* Indicator equal to one if student migrates out of district for education
* Indicator equal to one if student has applied to college
* Indicator equal to one if student has applied to college out of district
* Indicator equal to one if student has applied to scholarship or loan to finance their education

Index of hypothetical educational migration choices
* Likelihood of applying to colleges in out of district destinations
* WTP to stay within district (as measured by a conjoint experiment)
* WTP to stay within district (as measured by IRR exercise)
* Indicator equal to one if student plans to look for colleges in out of district destinations
* Subjective rating of out-of-district destinations

I will rely on inverse-covariance weighted Anderson indices to construct the primary outcomes. I will measure and analyze hypothetical choice outcomes for parents and students. I will (1) adjust for multiple hypothesis testing and (2) conduct a joint hypothesis test for the primary outcomes. In addition, I will test for heterogeneity with respect to gender, academic stream and baseline networks.

Secondary outcomes capture mechanisms shaping choices as described in primary outcomes, and fall into three groups: (i) beliefs about expected costs and benefits associated with different choices, (ii) constraints that limit the feasible choice set, and (iii) individual-specific traits that govern how students process costs and benefits.

Expected costs and benefits:
* Pecuniary outcomes (expected living costs and wages upon graduation)
* Non-pecuniary outcomes (expected happiness, safety, mental wellbeing, social ties, parental approval, freedom, effort and societal costs of moving)

Constraints:
* Efforts (search and study)
* Information (general information regarding college admissions and destination-specific information)

Traits:
* Gender norms index
* Generalized self-efficacy index
* Aspirations index

In addition, I will analyze (1) college quality, as defined by cutoff scores, placement packages and rates of colleges (depending on data availability); and (2) content of conversations between mentor-mentee pairs.

Students are randomly allocated to equally sized treatment and control groups, conditional on strata formed by grouping students in the same school and academic stream together.

Not available

Randomization done in office by a computer

Individual

No

0

2500 students

1250 students control, 1250 students treatment