On this page
Starting in late 2003, recruiters visited villages near Delhi to tell young women about call-centre and back-office jobs. Three years later, girls there were more likely to be in school. The study is Robert Jensen's "Do Labor Market Opportunities Affect Young Women's Work and Family Decisions? Experimental Evidence from India", published in The Quarterly Journal of Economics in 2012. Its central idea: when families learn that well-paid jobs for educated women exist and can be reached, they invest more in their daughters.
How was the BPO recruiting study designed?
It was a randomised trial. Jensen listed rural districts about 50 to 150 km from Delhi in Haryana, Punjab, Rajasthan and Uttar Pradesh, then randomly assigned 80 villages to get recruiters and 80 to serve as a comparison group (J-PAL summary). Because chance chose the villages, later differences can be credited to the recruiting.
Eight experienced Delhi recruiters ran four-to-six-hour sessions for women in each treatment village: jobs and pay, résumés, mock interviews and an English check, with a warning that jobs were not guaranteed. They held the session three times, a year apart, and offered free placement help in between (Jensen 2012).
The survey covered 20 random households per village, 3,211 in all, in 2003 and again in 2006. Starting BPO salaries in 2003 were often 5,000 to 10,000 rupees a month, about twice what similarly educated non-BPO workers earned, and the jobs typically needed 10 or 12 years of school (Jensen 2012).
What did the study find about girls' education?
The main numbers from the published paper, comparing the two groups in 2006:
Girls aged 6 to 17 were 5.0 percentage points more likely to be enrolled in school, against 76% in comparison villages, closing about 60% of the baseline boy-girl enrollment gap.
The gain was larger for girls aged 11 to 17 (5.9 points); for ages 6 to 10 it was 3.1 points and not statistically significant.
Girls aged 5 to 15 had higher BMI-for-age z-scores (up 0.24), which Jensen reads as better nutrition or health care. Height did not change.
Women aged 18 to 24 were 4.6 percentage points more likely to hold a BPO job and 2.8 points more likely to be enrolled in computer, English or similar training courses.
Boys and men showed no change in schooling, health or work.
What did it find about marriage and children?
Women who were 15 to 21 at baseline were 5.1 percentage points less likely to have married and 5.7 points less likely to have had a child by 2006, and they reported wanting about 0.35 fewer children, against an average of 3.0 in comparison villages (Jensen 2012).
Jensen is cautious. With three years of data, he cannot tell whether this is a delay or a lasting change, and desired family size is a stated intention, not an outcome. The data also cannot show who made these decisions: the young woman, her parents or, for children, her husband. These are changes in timing, not a judgement of any family's choices.
What does the study not show?
It did not create jobs. Recruiters connected women to existing jobs, and Jensen notes they may have got them at the expense of others (Jensen 2012).
It may not travel everywhere. These villages were larger and closer to a city than the average Indian village; Jensen would not generalise to all of rural India.
It covers one kind of work. BPO jobs are white-collar; Jensen would not necessarily extend this to farm or factory work.
It is one trial. We found no direct replication. Non-randomised studies point the same way: Oster and Millett (NBER, 2010) found the number of children enrolled rose about 5.7% at schools near a new IT-enabled service centre, driven entirely by English-language schools, and Heath and Mobarak (NBER, 2014) found that girls near garment factories in Bangladesh were more likely to be enrolled in school and delayed marriage and childbirth.
What can Indian families take from it?
This is our inference, not a finding, drawn from villages near Delhi in 2003 to 2006: what families believe about future jobs seems to shape how much they invest in a daughter's education.
Make careers concrete. Name three real jobs that need Class 12 maths or science and what they pay.
Talk to someone who does the job. A cousin or neighbour in that field can be specific about the work, pay and how to apply.
Map the path. For each job, list the subjects, exams and courses needed.
Pay attention in the secondary years. In the trial, the clearer school gain came for girls aged 11 to 17, when enrollment was lowest (Jensen 2012).
Read more: a trial that helped parents explain why maths and science matter, what ASER 2024 found about reading and maths in rural India and what the Teaching at the Right Level trials found.
If your daughter knows where she wants to go but not where to start, Fermi's AI tutor can work out what she already knows and build a plan for the session.



