Periodic Reporting for period 1 - EDINCLUSION (UNDERSTANDING AND IMPROVING SOCIAL INCLUSION IN INDIAN PRIVATE SCHOOLS)
Berichtszeitraum: 2023-02-01 bis 2025-07-31
Private providers account for a substantial share of education provision in many low- and middle-income countries (LMICs). In India, such private institutions account for ~30% of rural enrolment and 50% of urban enrolment at the primary school level. Since access to fee-charging private institutions is frequently constrained by parents’ ability-to-pay, this leads to concerns of social stratification.
EDINCLUSION proposes an ambitious research agenda focused on understanding and improving school integration in India, which is the world’s largest school system, and also among the most unequal. Specifically, the proposal has three broad aims: (i) to study the current effectiveness of quotas mandated by the Right to Education Act 2009 to expand access to private schools for children from disadvantaged backgrounds, (ii) to evaluate potential strategies to reduce application barriers for poorer households, and (iii) to study demand and quality variation in private school markets both in urban and rural areas. These questions are of substantial scientific and social importance for the study of private markets in education in low- and middle-income countries.
The research strategy is three-fold: (i) first, using “natural” lotteries embedded in the application system for allocating quota-applicants to school to establish the current effectiveness of the policy, (ii) using large randomized control trials to investigate the effectiveness of various targeting and bundling choices to improve application rates, (iii) using a combination of exogenous variation from (i) and (ii) along with methods from observational causal inference and to study variation in school quality and models of parental demand for specific schooling attributes. Each of these steps involves substantial methodological innovation in measurement and extensive primary data collection. This project is executed in partnership with state governments and Indus Action, a leading civil society organization; survey operations are managed by JPAL South Asia.
EDINCLUSION proposes an ambitious research agenda focused on understanding and improving school integration in India, which is the world’s largest school system, and also among the most unequal. Specifically, the proposal has three broad aims: (i) to study the current effectiveness of quotas mandated by the Right to Education Act 2009 to expand access to private schools for children from disadvantaged backgrounds, (ii) to evaluate potential strategies to reduce application barriers for poorer households, and (iii) to study demand and quality variation in private school markets both in urban and rural areas. These questions are of substantial scientific and social importance for the study of private markets in education in low- and middle-income countries.
The research strategy is three-fold: (i) first, using “natural” lotteries embedded in the application system for allocating quota-applicants to school to establish the current effectiveness of the policy, (ii) using large randomized control trials to investigate the effectiveness of various targeting and bundling choices to improve application rates, (iii) using a combination of exogenous variation from (i) and (ii) along with methods from observational causal inference and to study variation in school quality and models of parental demand for specific schooling attributes. Each of these steps involves substantial methodological innovation in measurement and extensive primary data collection. This project is executed in partnership with state governments and Indus Action, a leading civil society organization; survey operations are managed by JPAL South Asia.
EDINCLUSION built upon preliminary work that, at the time of writing the proposal, suggested that most applicants to the quota seats were inframarginal. Since that time, we have made substantial progress:
1. Completion of foundational research for ERC-funded fieldwork: As noted earlier, this proposal built on previous research and further work that was planned (but not implemented) at the time of application. An important part of our work in 2023 focused primarily on completion of the analysis of previously collected data and of ensuring that we took in all relevant feedback from scientific colleagues in designing the complex set of experiments that would be funded by ERC. The initial working paper was updated, submitted to successive journals and has now received a “Revise and Resubmit” decision at the Review of Economic Studies. This paper was also presented in several conferences and invited seminars at ~10 universities. A clearer understanding of past data, plus the feedback from colleagues globally, on the first paper was crucial in helping us better design the field experiment and measurement in the ERC-funded fieldwork.
2. Revised design for the field experiment to improve access: Based on the results from the foundational research, we revised the design of field experiments to improve application rates in the private school quotas. In particular, we broadened the scope of the experiment substantially to (a) study multiple potential ways to target potential applicants, (b) ensure that we are statistically well-powered both for detecting complementarities between information only and application assistance, and (c) generate the optimal data structure to then use machine learning techniques to improve targeting. We also revised the timeline of the activities to first prioritize the experiment and then the study of administrative lotteries. The final design required a substantial redesign of both sampling and measurement and is substantially improved (in terms of statistical precision and scope) than the initially proposed experiment. In particular, we made several innovations in measuring the preferences of parents for their children’s schooling and the constraints that prevent policy take-up, as well as in the design of experiments to study geographic targeting of interventions where reliable population rosters may not be available.
3. Completion of two large rounds of fieldwork in Chhattisgarh: We conducted baseline household surveys in Jan-March 2024 covering ~10,000 households in 7 districts. A subset of these households were randomized to also (i) receive information about the upcoming application round OR (ii) receive the offer of assistance to help apply OR both (to estimate if there are complementarities between the two treatments). These households were then reinterviewed between July-September 2024 to collect data on the primary outcomes i.e. whether they applied, whether they received a seat and the enrolment of all children in the households; we also administered age- and context-appropriate learning assessments to these students. A further round of data collection is planned from July 2025 to collect data that will allow us to estimate effects of school types on learning, and also to make substantive progress on understanding school markets.
4. Initial analysis of experimental results: Between November 2024 and March 2025, we have also finished the initial analysis of the experiment on all pre-specified outcomes (i.e. the “core” results of the experiment). We have now presented this in workshops in several universities internationally for initial feedback. The principal intention, beyond dissemination, is to also collect scientific feedback from some of the best research groups globally on further analyses that should be prioritized before proceeding with preparation of initial manuscripts.
5. Design of data collection for studying effects of natural lotteries: The basic structure of proposed data collection for Work Package 1 has not changed substantially, but has required substantial data cleaning and analysis in advance of the sampling for data collection using phone-based surveys. This process was completed in January 2025 and phone-based data collection has now started in March 2025. In person collection of data on student learning outcomes will then proceed from July 2025.
1. Completion of foundational research for ERC-funded fieldwork: As noted earlier, this proposal built on previous research and further work that was planned (but not implemented) at the time of application. An important part of our work in 2023 focused primarily on completion of the analysis of previously collected data and of ensuring that we took in all relevant feedback from scientific colleagues in designing the complex set of experiments that would be funded by ERC. The initial working paper was updated, submitted to successive journals and has now received a “Revise and Resubmit” decision at the Review of Economic Studies. This paper was also presented in several conferences and invited seminars at ~10 universities. A clearer understanding of past data, plus the feedback from colleagues globally, on the first paper was crucial in helping us better design the field experiment and measurement in the ERC-funded fieldwork.
2. Revised design for the field experiment to improve access: Based on the results from the foundational research, we revised the design of field experiments to improve application rates in the private school quotas. In particular, we broadened the scope of the experiment substantially to (a) study multiple potential ways to target potential applicants, (b) ensure that we are statistically well-powered both for detecting complementarities between information only and application assistance, and (c) generate the optimal data structure to then use machine learning techniques to improve targeting. We also revised the timeline of the activities to first prioritize the experiment and then the study of administrative lotteries. The final design required a substantial redesign of both sampling and measurement and is substantially improved (in terms of statistical precision and scope) than the initially proposed experiment. In particular, we made several innovations in measuring the preferences of parents for their children’s schooling and the constraints that prevent policy take-up, as well as in the design of experiments to study geographic targeting of interventions where reliable population rosters may not be available.
3. Completion of two large rounds of fieldwork in Chhattisgarh: We conducted baseline household surveys in Jan-March 2024 covering ~10,000 households in 7 districts. A subset of these households were randomized to also (i) receive information about the upcoming application round OR (ii) receive the offer of assistance to help apply OR both (to estimate if there are complementarities between the two treatments). These households were then reinterviewed between July-September 2024 to collect data on the primary outcomes i.e. whether they applied, whether they received a seat and the enrolment of all children in the households; we also administered age- and context-appropriate learning assessments to these students. A further round of data collection is planned from July 2025 to collect data that will allow us to estimate effects of school types on learning, and also to make substantive progress on understanding school markets.
4. Initial analysis of experimental results: Between November 2024 and March 2025, we have also finished the initial analysis of the experiment on all pre-specified outcomes (i.e. the “core” results of the experiment). We have now presented this in workshops in several universities internationally for initial feedback. The principal intention, beyond dissemination, is to also collect scientific feedback from some of the best research groups globally on further analyses that should be prioritized before proceeding with preparation of initial manuscripts.
5. Design of data collection for studying effects of natural lotteries: The basic structure of proposed data collection for Work Package 1 has not changed substantially, but has required substantial data cleaning and analysis in advance of the sampling for data collection using phone-based surveys. This process was completed in January 2025 and phone-based data collection has now started in March 2025. In person collection of data on student learning outcomes will then proceed from July 2025.
The foundational work described above is already at an advanced stage and promises to be the most comprehensive investigation of the current effectiveness of India’s Right to Education Act – the largest voucher-like mechanism for private primary schooling globally – and important also for the broader literature on barriers to application for welfare programs in developing countries.
Preliminary results from the ERC-funded field experiments indicate substantial further progress on understanding how effectiveness could be enhanced. Beyond the scientific impact, we expect also that these will lead to potential policy improvements and, once results are finalized, they will also be shared and discussed with the Government of Chhattisgarh and Indus Action. In the second half of the grant period, we hope to also take insights from the work done until 2025 in Chhattisgarh to inform policy discussions in another state to assess external validity.
Preliminary results from the ERC-funded field experiments indicate substantial further progress on understanding how effectiveness could be enhanced. Beyond the scientific impact, we expect also that these will lead to potential policy improvements and, once results are finalized, they will also be shared and discussed with the Government of Chhattisgarh and Indus Action. In the second half of the grant period, we hope to also take insights from the work done until 2025 in Chhattisgarh to inform policy discussions in another state to assess external validity.