Cameras improve teacher attendance

Teacher absence ranges from 20% to over 50% in different states, and makes a mockery of free and universal education. No wonder children drop out in droves, and functional illiteracy is high even for those who complete school. In such circumstances, the government’s plan to double spending on education will simply double the waste.

One possible solution comes from Sewa Mandir, an NGO, whose experiment has been analysed in a research paper by two American scholars (Monitoring Works: Getting Teachers to Come to School, by Esther Duflo of MIT and Rema Hagner of New York University)

Sewa Mandir runs non-formal schools in hilly, scattered villages of Udaipur District. Years ago it sanctioned the firing of persistently absent teachers, but was unable to monitor their absence. So, as an experiment in 2003, it equipped 60 schools with cameras having a tamper-proof time-and-date function. Each teacher had to ask a student to take a photo of himself/herself along with at least eight other students at the start and end of school, which had to be at least five hours apart.

The teaching record of these camera schools was then compared with that of 60 other normal schools in the neighbourhood. In normal schools, teachers earned a flat Rs 1,000/month. In camera schools, teachers got a base salary of Rs 1,000 for 21 days a month; a bonus of Rs 50 for every extra day worked; and a fine of Rs 50 for every day absent (maximum fine Rs 500). So, pay in camera schools ranged from Rs 500 to Rs 1,300/month.

Result? Over 18 months, the camera schools recorded teacher absence of 22%, against 42% for normal schools (and 44 % for all schools before the experiment). Hence cameras almost halved teacher absence. Teachers were present over 90% of the time in 35% of camera schools, against just 1% of normal schools. Best of all, random checks suggested that teacher presence remained high in the camera schools even after the experiment ended: the mid-set had changed.

Teachers were not hostile. On feedback forms, many said the program had instilled a new sense of discipline that they liked, apart from linking performance to pay. Indeed, teachers said that the cameras enabled them to better resist pressures from locally elites to do tasks other than teaching. But some complained that kids might arrive too late for the morning photo.

Children in camera schools received 10% more teaching time (or 34 more days per year) than in normal schools. Did this improve learning outcomes? Tests before and one year after the programme started showed that children in camera schools scored significantly higher (0.17 standard deviations) than in normal schools, and were 40% more likely to be admitted later into regular government schools. The sharpest improvement was recorded by children with higher initial scores: the impact was negligible for the bottom half of students.

Could the data have been fiddled? Teachers could in theory have collected children in the morning and evening to pose for photos, without teaching in between. But such manipulation would be difficult in scattered habitations where students cannot easily be gathered together at any one point and time. Again, teachers might in theory attend on more days without actually teaching more. But the test scores suggest improved teaching.

The costs of cameras, film and developing charges, and supervision averaged Rs 5,379/school/year, or 40% of a teacher’s annual salary. That sounds high, yet costs only 40% as much as a rival strategy that yielded similar learning gains, hiring a second teacher per school. Today, cheap digital cameras are available, requiring no film or developing costs, and that will cut costs further.

The main lessons are that teachers respond to rewards and fines; closer monitoring improves discipline and morale; and learning outcomes can be improved without spending vast sums on additional teachers and school buildings. More important, the same technique can be applied to improve attendance in primary health centres.

The strategy may not work in urban areas where teachers can more easily get children together in the morning and evening to fake attendance. It may not be needed for single schools that can be monitored directly by their owner or NGO. But clearly cameras could play a role in monitoring chains of schools run by owners, NGO, or zila parishads. Such chains now provide a significant share of Indian education.

What about government schools, which dominate education? Teacher unions will be outraged at the linking of performance to pay: they have never seen any connection between the two. No political party seems willing to confront teachers’ unions.

Yet zila parishads are increasingly being allowed to appoint teachers, and they could use cameras to monitor attendance. This is not a panacea for all educational ills, or a reason to abandon other experiments such as education vouchers. But it is one way forward.

What do you think?