Food for thought: On the design of school subsidy programmes

Despite significant increase in primary school enrollments, student attendance rates are less than 70% in public schools. This column argues that India needs to start evaluating its existing school subsidies systematically. It finds that provision of free cooked meals at schools that are sufficient not just in terms of quantity but also quality and variety will ensure better targeting and help get more children in school.

Despite significant improvements in school enrolment in recent years, almost 30% of Indian children of primary school age still continue to be out of school, while the daily school attendance rate of those enrolled is less than 70% (National Family Health Survey 2005-06 and Government of India Report 2006-2007). Concerns about the poor quality of public schools notwithstanding, ensuring that children attend school regularly and continue with their education remains a key challenge for India.

Governments in India and elsewhere have tried to increase school participation through policies that subsidise the cost of education, in some cases directly and in others indirectly, such as providing school meals, scholarships, free textbooks and uniforms. Yet while these efforts have been popular policy tools for increasing school participation, the tight budgets in developing countries make it vitally important that governments evaluate whether these programmes are successful. Surprisingly, insights into how programme benefits vary by the design of school transfer programmes are fairly limited, if not absent.

Making a meal of it

In a study of India’s school meal programme, the largest such scheme in the world, I assess the effect of transition from the monthly distribution of free food grains to the daily provision of free cooked meals to school children on enrollments and attendance in a rural area of India in 2003. I find that there could be gains from shifting from the ‘off-site’ programme of distributing food grains to the ‘on-site’ program of providing daily cooked meals on school premises (Afridi 2010).

To isolate the effect, I compare the difference in grade level participation rates between schools that began serving cooked meals to the participation rates of a control group of schools whose programme status did not change during this period The study reveals that the switch to free meals at school significantly improved daily attendance among younger children. The average monthly attendance of girls in grade 1 was more than 12 percentage points higher in treatment schools while there was a positive but statistically insignificant effect on grade 1 boys’ attendance. As expected, there was no effect on enrolment due to this switch to free meals, given that the increase in the implicit value of the subsidy due to conversion of foodgrains into a cooked meals was nominal.

Half-baked policies

The findings suggest that while an ‘off-site’ school meal programme allowed for sharing of the food within the family, an ‘on-site’ cooked meals programme was better targeted towards the children attending school, reducing the possibility that children who weren’t enrolled could still receive free meals if their siblings attended school. So even though the implicit value of the transfer may not have been significantly different between the two types of programmes, the incentive to attend school regularly was higher in the on-site scheme, particularly for girls. Furthermore, the off-site school meals programme was designed to be conditional on a minimum monthly attendance but such conditions are rarely imposed by public schools.

In an on-going study with colleagues (Afridi et al. 2012), I extend the investigation of the design of school meal programmes to an urban context in India. Public primary schools in the municipality of Delhi moved from an on-site programme of providing packaged, ready-to-eat snacks to serving cooked meals to all enrolled students in 2003. This change in programme design led to less portability but more variety in the food that was served to children. The change in the design of the on-site meal programme was rolled out in two phases across all public primary schools over a period of five months in the same academic year (July to December 2003). We compare the attendance rates of students in schools which transitioned early to those that transitioned late to cooked meals using data for the period September 2002 (ready to eat regime) to February 2004 (cooked meals regime).

We find that on average the cooked meal programme resulted in a 2 to 3 percentage point increase in students' monthly attendance. This suggests that just by modifying the nature of a programme, with a marginal increase in cost per child, the programme benefits were markedly higher.

For the same sample of municipal schools in this study there was a second change in the school meal programme - the introduction of better quality cooked meals in 2004-2005 academic year. The Municipal Corporation of Delhi (MCD) began evaluating the quality of food provided by contractual meal providers in 2003 and discontinued contracts of those found to be below par. Schools that were earlier being served by 'low quality" providers were now assigned ‘high quality’ providers. We find that an improvement in the quality of the school meals significantly raised students’ monthly attendance, over and above the improvement in participation due to the shift from ready-to-eat to cooked meals. The above research underlines the need for closer attention to the design and quality of school transfer programs in order to yield higher returns within constrained budgets. For instance, substantive research suggests that greater resources in the hands of mothers of targeted children may result in a higher proportion of household resources sticking to the intended beneficiary than, say, transfers to the father (Luke and Munshi 2011, Qian 2008). If this is true, it may be worthwhile to study whether the benefits of such targeting of intended beneficiaries outweigh any additional costs induced due to it.

Informing policy through rigorous evaluations of the relative effectiveness of varied designs of school subsidies are essential in order to get the highest possible program benefits per rupee spent. In general, this strengthens the case for carefully assessing the efficacy of variations in design within the administration of public programs, for instance through randomised variations, before universal rollout.