Does more schooling infrastructure affect literacy?

ABSTRACT

This paper examines how the expansion in schooling infrastructure of girls as part of India’s Education for All program has increased female literacy and reduced gender gaps. To identify causal effects, I exploit the variation according to the targeting scheme of the programme which involved classifying subdistricts as either educationally backward or not. Using a regression discontinuity method, I find significant expansion in the number of girls’ schools and residential schools for girls, but no significant positive effect on either female literacy or the gender literacy gap. Cost-effective methods other than an untargeted, large-scale infrastructure programme should be explored.

KEYWORDS:

• Schools
• education
• literacy
• gender
• Sarva Sikshya Abhiyan

Acknowledgement

I am grateful to Mark Borgschulte, Tatyana Deryugina, Marieke Kleemans, Benjamin M. Marx, Adam Osman, Elizabeth Powers, Rebecca Thornton for their helpful comments and guidance. I thank the seminar participants in the applied micro research lunch talk, participants at the ISI (Delhi), NARSC, and DSE winter school conference for their useful feedback. I appreciate the efforts of the DISE team including Rohit Hans, Aparna Mookherjee, Arun Mehta for helping with the data and answering my queries. Clarifications and information provided by Pratibha Kumari and Anil Kumar from the Census division of India are highly appreciated. Any remaining errors are my own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Blocks – also known as tehsils, mandals, or subdistricts – are the subdivisions of a district in India. The gender gap in literacy rates is defined as the difference between the male and female literacy rates.

2. KGBV was launched as a separate scheme in 2004 but was merged with SSA in April 2007.

3. The initial release of funds for the scheme was Rs 584 crore ($123.75 million) in the school year 2001–2, which increased more than ten times to 6,846 crore ($1,510.6 billion) by 2004–5 and to 30,793 crore ($6.73 billion) by 2010–11.

4. In conducting the census, the head of the household (or other respondent) is asked for the number of literate persons in their household. Sometimes, the interviewer asks the household members to read. The literacy rate in India is calculated by taking the percentage of literate persons above age six in the population.

5. More details on the matching procedure are available in the data appendix. The proportion of blocks classified as EBB remains similar in the reduced sample, and the drop in observations does not seem to be systematic. Also, the above drop is for the entire sample, and for this reason the RD estimates, which consider observations only around the cut-off, must not be biased significantly.

6. To my knowledge, the criteria used to classify blocks as educationally backward are not used for any other classification or for implementing any other programmes.

7. The alternative of restricting the sample to ones which satisfy the RFLR threshold and using gender gap in total literacy rate as the assignment variable leads to a smaller sample. However, the results are not different as seen in Figure A.1 of the appendix.

8. There were some blocks in districts in which at least 5% of the population was SC or ST or in which the female literacy rate of the SC or ST group was below 10%, which received NPEGEL, even if they did not satisfy the general criteria. Additionally, some selected urban slums also received the NPEGEL. Additional details in data appendix. Source: Ministry of Human Resource and Development framework for implementation of SSA, MHRD (2009)

9. For a detailed discussion and review, see Lee and Lemieux (2010).

10. The states and union territories of Delhi, Chandigarh, Goa, Andaman and Nicobar Islands, Sikkim, Lakshadweep, Puducherry, Daman and Diu, Dadra, and Nagar Haveli included no blocks classified as educationally backward. I also conduct the analysis excluding states which did not have any block classified as EBB; the results remain similar.

11. There were around fifteen schools built in the decade before the implementation of SSA (and KGBV scheme), which were classified as KGBV in the data (observed in the plot).

12. The results remain similar for total literacy rate and total female literacy rate as well.

13. For example, the 2011 Annual Status of Education Report (ASER) for rural India states that 38.4% of students in grade 1 could not recognise letters and 36.5% could not recognise numbers (ASER 2011). Banerjee et al. (2010), in their experiment in Jaunpur District, find similar evidence. Muralidharan (2017) suggests that improving school inputs in the ‘business as usual’ manner will have little effect on learning outcomes; he recommends changes in pedagogy and governance.

14. A similar concern exists for other efforts which involve building infrastructure to solve social problems, such as the Total Sanitation Campaign in India, which aims to eliminate open defecation. The campaign included construction of toilets throughout the nation, but that guaranteed neither that the toilets would properly function nor that citizens would use them.