The State and the Market in Education Delivery

The State and the Market in Education Delivery: Implications for Implementing the RTE Act Karthik Muralidharan University of California - San Diego; NBER; J-PAL School Choice National Conference, New Delhi 16 December, 2009 1

Background • Central role for education in the uplifting of historically disadvantaged communities in all countries • Especially relevant in Indian context, where the rigidities of caste perpetuated inequalities in education access across generations, making improved education access a critical component of social policy • As recently as 1971, the data shows substantial under-provision of public goods (schools, clinics, roads, electricity) in areas with higher scheduled caste and tribe populations (Banerjee and Somanathan, 2006) • But combination of political empowerment (especially of scheduled castes) and policy focus on universal provision has substantially reduced inequalities in access to primary education • Sarva Shiksha Abhiyan has been a success in terms of access and enrollment with over 95% of children enrolled in school 2

The Challenge of Education Quality • However, successes on the quantity front mask severe problems in education quality • ~60% of children aged 6 -14 in India cannot read a simple paragraph, though over 95% enrolled in school (PRATHAM, 2008) • Research shows that the returns to education (both at the individual and aggregate levels) are driven more by the quality than the quantity of education (Hanushek and Woesmann, 2008) • The lack of education quality severely limits the ability of education to serve as a vehicle of social mobility • As the locus of job creation moves to the private sector from the public sector, the premium is on skills as opposed to paper qualifications • Most policy discussions on quality of education focus on increasing public spending, but very little on improving effectiveness of the money that is spent • Teacher accountability and effectiveness is the central issue (~90% of recurrent spending goes to teacher salaries) 3

All India Teacher Absence Map (Public Schools) State Maharashtra Gujarat Madhya Pradesh Kerala Himachal Pradesh Tamil Nadu Haryana Karnataka Orissa Rajasthan West Bengal Andhra Pradesh Uttar Pradesh Chhatisgarh Uttaranchal Assam Punjab Bihar Jharkhand All India Teacher Absence (%) 14. 6 17. 0 17. 6 21. 2 21. 3 21. 7 23. 4 23. 7 24. 7 25. 3 26. 3 30. 6 32. 8 33. 8 34. 4 37. 8 41. 9 25. 2% 4 Source: "Teacher Absence in India: A Snapshot" (Michael Kremer, Karthik Muralidharan, Nazmul Chaudhury, Jeffrey Hammer, Halsey Rogers), Journal of the European Economic Association, vol. 3, no. 2 -3, April-May 2005, pp. 658 -67

The Emergence of Private Schools i. One response to the lack of performance of government schools has been a near explosion of private schools in the past several years i~50% of children in urban India and 20% of children in rural India attend fee-charging private schools (IHDS 2005, ASER 2008) i. These are not the fancy private schools that people in urban areas typically think about, but rather “budget” private schools that cater to poor and lower middle class parents (Muralidharan & Kremer, 2007; Tooley 2009, etc. ) i. Main sources of competitive advantage are: i Higher accountability (~175 times more likely to fire absent teachers) i Flexibility/Responsiveness to what parents want (Eg. teaching English early) i Significantly higher student outcomes (attendance, test scores) even after controlling for observable differences in family background i Much more cost effective (smaller classes, less multi-grade teaching, though spending/child is 3 -4 times lower than spending in government schools) i The critical driver of private school economics is lower teacher salaries (which allows hiring many more teachers), and better accountability 5

Salary Distribution by School and Teacher Type Source: Calculated from data collected as part of ongoing research by author in rural Andhra Pradesh

Public School Failure and Private School Entry 7 Source: “Public and Private Schools in Rural India” by Karthik Muralidharan and Michael Kremer in School Choice International, MIT Press, 2008

Exit versus Voice i. People who are dissatisfied with a relationship can try to improve their outcomes either through exit or voice (Hirshman, 1972) i. To a large extent (but not completely), accountability in markets works through exit (and the threat of it), while accountability in political spaces work through voice i The concepts are inter-linked in that increasing the power to exit, can also increase voice because it is more credible i. Schools are interesting because they combine elements of market (possibility of moving to private schools) and political spaces (community control and collective action) i. So attempts to empower disadvantaged communities should either strengthen voice or enable exit (or both) i Assumes that demand is not a constraint (and this appears to be the case) i. Traditional thinking on improving accountability has focused much more on voice (decentralization, parent-teacher associations, etc) 8

Evidence on the Effectiveness of Voice i. Mixed evidence on PTA’s i Correlations show that the existence of PTA’s has no relation with teacher absence, though a more active PTA is correlated with lower teacher absence i Suggests limited effectiveness of a supply-side initiative to create PTA’s i. No evidence to suggest decentralization reduces absence (to date) i We use NUEPA handbooks to code an index of education decentralization (planning, information management, etc) i Find no relationship with teacher absence (at least with current forms of decentralization; may be better with more complete devolvement) i. Mixed evidence on the impact of information on teacher absence i Banerjee et al. find no impact of providing more information to Village Education Committees in UP regarding their powers and quality of education i Pandey et al. do find positive effects of an information intervention in MP i. Key challenges to the effectiveness of voice are: i Limited authority of communities over teachers i Even if the authority were vested in the communities, there would be a need 9 for effective collective action, and risks of local elite capture would remain

Voice in the face of Elite Exit i. The effectiveness of “voice” depends on the power of those exercising it i. The challenge for schooling is that the “voice” channel is progressively weakened by the exit of the elite from being recipients of public schooling (true for other public services as well) i. In our all-India sample (collected in 2003): i Over 80% of govt. school teachers send their own children to private schools i In villages with a private school, members of the Gram Panchayat are around 20% more likely to send their own kids to the private school i But absence rates are around 17% lower if all GP members had their own kids in the government school i. Of course, the direction of causation might be the opposite i. The main point is that of “multiple equilibria” i In one, the govt. school is good and every one sends their kids there; voice is strong; quality is maintained i In the other, the govt. school is weak; private schools show up; elites migrate 10 there; poor have no exit options and limited voice (worst of both worlds)

Enrolment and Outcomes by School Type i. Our most recent data (collected across 5 districts in Andhra Pradesh in 2008) show that in villages with a private school, the patterns of enrollment by school type and caste are: i General: 31% Govt. ; 69% Private i OBC: 39% Govt. ; 61% Private i SC: 69% Govt. ; 31% Private i. Not surprising given that private schools charge fees (typically around Rs. 150 -200/month) i. Correlations show that the learning levels in the private schools are significantly higher (0. 75 SD) than those in private schools i These are not causal estimates and could reflect unobserved variables i But the “private school” effect persists even after controlling for school and household characteristics i. Process indicators also suggest the superiority of private schools i. Most revealing indicator is parental revealed preference i What does it say about the quality of the product (govt. schools) if you cannot even give it away for free? Even with a positive subsidy! 11

Combining the Best of Both Worlds? i. The strength of markets for schools (with competition among suppliers and basic regulation on safety etc. ) are that they are: i Customer-centric, flexible, and responsive i Accountable i More likely to engender innovation i. But the main weakness is that the market does not care for you if you don’t have purchasing power i. The idea that governments have to provide education, health, etc is based not just on “public good” considerations, but also on the notion that it is the only way to ensure universal access i. Government provision typically does do a better job of providing universal access, but severe problems on the dimensions above i. The idea of school vouchers/scholarships: i Fund students and not schools i Parents can choose any school they like – public or private (subject to basic regulation), pay with a voucher, and the school is reimbursed directly by the 12 government (can empanel eligible private schools for quality control)

Implications for Empowerment i. The point is not to claim that all private schools are superior to govt. schools (this is not the case) i. Rather, the point is that the disadvantaged should have the same capacity to exercise choice as those who are better off i. Why should disadvantaged groups be subject to the vagaries of state provision, when nearly anyone who can afford to secede to the private sector chooses to do so? i. Vouchers are a more powerful tool than reservations, because they can benefit the entire community as opposed to the few who manage to secure seats in reserved categories i. Vouchers are also a much more flexible instrument because the value can be calibrated to account for the extent of disadvantage i. Much more likely to engender innovative supply responses i. Relevant to all levels of education 13

Concerns about Vouchers/Choice i. Are private schools really the solution? i Family differences could be the main driver of differences with govt. schools i Private schools are also “club” goods who market themselves by their exclusivity and by who they exclude as much as who they include i Many schools may exclude children from disadvantaged communities even with a voucher i. Can poor/illiterate parents make well informed schooling choices? i. Will it lead to fraudulent enrollment and kids not going to school? i. Could it lead to a balkanization of schools along ideological/ religious/ethnic lines? i Schools are also about producing a shared civic identity and not just knowledge and skills i. Does it mean giving up on the public system and the government abdicating responsibility for education? i. Let us address each of these concerns 14

Summary i. Aim of this talk has not been to promote private schools, but to present the facts on public and private schools and to provoke thinking about alternative means to empowerment beyond state provision i. The economic case for vouchers is not based on “public” vs. “private” but rather on the idea that increased competition and choice will improve both public and private schools i. But the most compelling reason to consider the idea is on grounds of equity and social justice i. Vouchers are not a panacea to the problem of quality education (many private schools are also bad), but can provide disadvantaged communities choices similar to those available to better off groups i. Several practical questions need to be answered – but these are empirical and not theoretical questions and are best addressed by conducting small pilot projects with rigorous impact evaluation i. The 25% provision in the RTE Act provides an opportunity for such pilots i. We are conducting one such pilot project in Andhra Pradesh with Azim Premji Foundation to study the impact of such a scholarship program 15 i Designed to be directly relevant to the implementation of the 25% provision

Design of AP School Choice Pilot and Study i Aim is to provide poor children in rural AP with increased choice to attend a school of their choosing by providing a scholarship to cover the fees of private schools i Baseline test conducted for all UKG and Class 1 children in ~180 villages in 5 districts of AP i Parents of children who took the assessment and had never been in a private school were invited to apply for the scholarships i All fees, books, uniforms, materials covered (paid directly to schools); meals and transport were not covered (average scholarship value of Rs. 3, 000/year which is ~25% of per child spending in govt. schools) i Scholarships can only be used in recognized private schools i Excess demand for scholarships, allocated by lottery i Schools cannot pick and choose “better kids”; have to accept all or none of the lottery winners (APF signs agreement with all participating schools - participation is completely voluntary) i Annual assessments at the end of each school year i Allocation of scholarship by lottery allows us to estimate the impact of private schools in “value addition” without being confounded by other variables i The process also generates a lot of learning for scale up and for figuring out the 16 operating principles by which this component of the RTE can be notified

Implications for RTE Notification i. The idea of leveraging private schools does not take away the relevance of using “voice” to improve the quality of govt. schools – especially on issues of teacher performance and accountability i. But, it is a “long route” to accountability that will take a longer time, while every passing year is another cohort that has experienced poor quality education i. Telling quote from Sarpanch in MP (an SC himself) on why he sent his children to the private school instead of trying to improve the govt. school: “Jab tak main ye school ko sudhaar sakoon, tab tak mere bachchon ka bhavishya hamesha ke liye bigad gaya hoga” i. Expanding private schooling options (through the 25% clause of the RTE) has the potential to empower entire disadvantaged communities as opposed to the select few who benefit from the status quo policies to promote social justice i. Well worth thinking hard about the details of how this will be implemented in practice and how it will be evaluated 17