FIXING SCHOOLS

The Problems

1. The quality of education in India is very low. The majority of students who have technically completed primary school are neither literate nor numerate.

2. Within a single school class, students can be of vastly different levels of achievement – and this inequality increases as you move through school.

3. An emphasis on ‘passing exams’ has meant that students focus on memorization rather than on picking up useful skills and concepts. Vocational training is not considered important.

4. The Right to Education has expanded access – but at the cost of forcing many private schools to shut down.

Reforming the Indian School Education System*

Karthik Muralidharan

A fundamental source of tension in the design of an education system comes from the fact that, historically, education systems have served two very different purposes.

First, they have sought to impart knowledge, skills and shared concepts of identity and citizenship. We can broadly refer to these as the ‘human development’ role of education.

However, education systems have also served a second purpose, which is to assess and classify students on the basis of educational ability and achievement and to select higher achieving students for higher education and occupations that aim to attract the most meritorious students. We can refer to this as the ‘sorting’ function of education.

A quick summary of Indian education history after Independence would suggest that the education system for the most part is driven by sorting rather than human development. Indeed, the Indian education system in its current form is perhaps best understood as a ‘filtration’ system rather than an ‘education system’.

Using this framework helps make sense of several deep structural challenges of the Indian education system that have been widely remarked upon.

1. There are massive inequalities in the overall education system – which on the one hand routinely produces students who go on to achieve global excellence in their fields and on the other hand produces the world’s largest number of primary-school-completing students who are not functionally literate and numerate at even a second- or third-grade level.

a. A major reason for this is that the syllabus and textbooks have not changed from a time when a much smaller fraction of students were in the education system. Since the focus of the system continues to be on ‘passing’ exams linked to the syllabus, children who fall behind the curriculum often end up learning close to nothing despite attending school.¹

b. This is because teachers, parents and students are not rewarded for improving learning at levels below the current grade level, because it will not help in passing the grade-level exam – and so students who fall behind early are left behind in perpetuity.

2. An obsessive focus on exams and marks has led to an education system characterized by rote learning to pass exams (often through cramming of past exam papers) as opposed to conceptual understanding that can be applied and used in practical situations.

a. This is also in part a consequence of point (1), because the only viable strategy for students who fall behind is to cram for exams and hope to pass them by having memorized questions that may appear in the exam.

b. Conceptual understanding is not prioritized as it is not rewarded by the exam system.

3. Very low levels of practical skills even among students who have notionally ‘passed’ various exams and possess various levels of paper qualifications.

Note that there is nothing wrong with sorting per se. Every society around the world aims to identify its most talented citizens and match them into leadership roles and occupations that affect society as a whole. It is also completely rational both for institutions of higher learning and employers to seek credible signals about the level of learning in preparation of students and for students to seek to credibly provide the signals.²

Further, sorting-based education systems may well have been efficient for agrarian societies where the fraction of knowledge-based jobs was small and where the economic and social returns to education were limited for those in agrarian and manual labour.

However, the problem with the sorting paradigm of education is that children who fall behind (overambitious) grade-level standards do not get a meaningful education because there is no provision for ‘teaching at the right level’. This has led to a massive waste of both time and money. Money is spent on building schools and hiring teachers, and effort is focused on keeping children in school (to prevent ‘dropout’), but very little actual learning is taking place (see evidence reviewed in Muralidharan [2013] for details).

The central design challenge for the Indian education system is that it was designed for sorting and not for human development. Almost every structural weakness of the Indian education system can be explained by this framework.

Turn to Appendix A for an illustration of the scale of this challenge. Indeed, perhaps the most important graphs to understand school education in India are provided in Appendix A based on data from Delhi and Rajasthan.

Thus, a fundamental goal for Indian education policy has to be to move the education system from a sorting and selection paradigm to a human development paradigm, one that will empower every citizen to be educated enough to have the foundation needed for a lifetime of continuous learning in whichever area of skill he or she may seek to learn.

The core building blocks for achieving this goal include:

1. Curriculum reform to reduce content and emphasize understanding.

2. Exam reform to provide both ‘absolute’ and ‘relative’ credentials.

3. A national mission to achieve universal functional literacy and numeracy by grade three.

4. Universal preschool education to support school readiness before grade one.

5. Reform of teacher training to emphasize pedagogy over theory.

6. Having greater clarity on the role of the state and the market in providing education in India – and both leveraging the private sector to achieve India’s education goals and regulating it adequately as needed (this requires reforms to the RTE Act).

7. Creating high-quality vocational education streams in school and integrating these with practicum-based training programmes for vocational education (in grades seven to twelve).

It is also essential for the education policy to emphasize certain key cross-cutting themes that have been neglected for the most part in education policy discourse in India – which so far has focused more on ‘what’ to do rather than on ‘how’ to do it in a manner that is compatible with fiscal and capacity constraints. These include:

1. Using evidence and research to better inform policy choices.

2. Cost-effectiveness.

3. Governance.

The cross-cutting themes are discussed first, followed by a more elaborate discussion of the seven building blocks identified above:

Themes

1. Using evidence and research to better inform policy choices: The last fifteen years have seen a sharp increase in the quality of evidence available on the effectiveness of various policies at improving education outcomes. Yet, this evidence is typically not reflected in the ‘business as usual’ policy choices that are made by the Union and state governments. Thus, while education policy makers should definitely conduct extensive consultations, it is essential to put more weight on recommendations backed by high-quality evidence.

2. Cost-effectiveness: A second key theme is that of cost-effectiveness of policy recommendations. Prior education policy exercises have been strong in articulating visions for education, but have usually paid less attention to the details of how this vision can be achieved keeping in consideration the financial and administrative constraints within which policies have to be framed and implemented. Given the combination of fiscal constraints and strong needs for spending in several sectors, a responsible policy exercise has to consider the cost-effectiveness of various policy options to achieve a given set of goals.

In particular, the evidence from high-quality studies above suggests that it may be possible to achieve sharp improvements in education outcomes even without increasing spending – by reallocating expenditure from less to more cost-effective interventions and policies.

It is worth paying special attention to the issue of teacher salaries, which comprise the largest component of education spending. Several studies have shown that there is no correlation between the level of teacher salaries and their effectiveness at improving learning outcomes. Yet, it is likely that many states will implement the Seventh Pay Commission award, which will absorb the major part of any increase in education spending in the coming decade, while being highly unlikely to improve learning outcomes.³

Thus, it is essential for the Union government to strongly urge state governments to innovate, evaluate cost-effective options for improving education outcomes, and use the fiscal and policy space provided by the Fourteenth Finance Commission award to try to pivot education expenditure from less to more cost-effective policies.

3. Governance: A last cross-cutting theme that is central to achieving the goals of the new education policy is governance. Studies in the past decade have identified shocking weaknesses in the education governance in the country. Thus, India has, for example, high rates of teacher absenteeism, as well as high rates of vacancies in key supervisory positions including block and district education officers. These officials are also transferred frequently (their average tenure is less than a year), which makes effective governance very difficult.

Here we recommend that the national education policy follow the guidelines established by the NITI Aayog under its School Education Quality Index initiative.

The goal of the state-level SEQI is to institutionalize a focus on improving education outcomes (learning, access, equity) as the principal aim of school education policy in India. It also recognizes that school education is primarily a state subject and aims to encourage state-level leadership in improving outcomes in cost-effective ways.

It is hoped that the annual calculation and dissemination of the SEQI, with a focus on measuring and highlighting the annual improvement of states, will:

a. Shift the policy focus to outcomes rather than inputs and programmes.

b. Encourage state-led innovation in cost-effective policy approaches to improve outcomes.

c. Facilitate documentation and sharing of best practices across states.

The SEQI has developed a series of governance indicators that states will be measured and ranked on, and against which progress will be tracked over time. These indicators reflect a high-level consensus on key governance indicators, and we urge the Ministry of Human Resource Development and the NITI Aayog to monitor progress on these.

We now turn to discussing the substantive topics.

1. Curriculum reform to reduce content and emphasize understanding: There is already a lot of good content and ideas in the Yash Pal Committee report on the New Curricular Framework that has not been acted on. So this section will be brief.

The current curriculum has way too much content, which in turn pushes the education system to deliver rote learning as opposed to conceptual understanding. A content-heavy approach is again consistent with a sorting system since the sheer volume of material to be covered makes it more likely that students with better natural academic aptitude (as opposed to better ‘educated’ students) will score better on exams.⁴

In an age when factual information is easily available on the Internet (which in turn is increasingly widely available through smartphones), the premium on memorization and regurgitation of facts is falling sharply relative to the ability to be able to ask and answer relevant questions by finding facts and opinions from different sources, assessing their relative merit and being able to effectively synthesize such content for better understanding and decision-making.

Thus, a core goal of education policy should be to simplify and narrow down the volume of topics covered and emphasize instead the ability to understand concepts, connect concepts across topics and apply them creatively in solving newer problems.⁵

2. Exam reform to provide both ‘absolute’ and ‘relative’ credentials: It is important for education reformers to recognize that despite the best intentions of educationists and national leaders to have education systems reflect a broader set of goals, the single most important determinant of what teachers, parents and students work towards is the content and structure of the exam system. Indeed, the examination system is the proverbial tail that wags the dog of the entire education system.

Thus, reorienting the education system towards a human development paradigm as opposed to a sorting paradigm will require a fundamental rethinking of the examination system in India, which in its present form is almost completely oriented towards the sorting function. As explained above, it is not viable in practice to wish away the sorting function of an education system. The problem with the status quo, however, is that:

a. Improvements in learning that are at levels significantly below the examination threshold are completely undervalued by all stakeholders (parents, students and teachers) because such improvements will not be seen in the current exam system.⁶

b. Thus, the default of instruction in the classroom pretty much follows the textbook and syllabus – regardless of the fact that the vast majority of students (at least in government schools) are way behind the curricular standards of the syllabus (see Appendix A).

c. As a result, students who fall behind grade-level standards before completing eighth grade typically learn very little in class. Their only hope of managing with the expectations of school is to attend coaching classes, cram from past test papers and hope to somehow pass exams (typically with very little understanding of the content).

d. While the labour market cares about actual skills, there is no credible way of learning about a student’s ‘absolute’ level of understanding/mastery of a concept. This is because the marks on a grade-level exam mainly serve to ‘rank’ students for admission to higher levels of education and do not convey any information to parents, students or employers about the absolute level of mastery of any relevant skills.

Thus, part of key reforms must be a national testing agency that can prescribe standards for ‘absolute’ levels of proficiency at topics, which are broken down at a much more granular level by subject and grade (similar to how software-based learning applications like Khan Academy and Mindspark are structured).

The availability of modular assessments that are organized in ascending order of skills will provide a critical source of feedback to parents, teachers, students and employers about the absolute competence that a student has attained and about the progress made at regular intervals (say, every month to three months).

While exams will continue to serve a sorting function based on student rank, the goal is for the education system to focus on absolute progress made by each student relative to his/her own prior level (regardless of the rank of the student in class and the extent to which he/she is behind the grade-level standards).

Thus, the nature of assessments will need to shift from saying ‘Student X scored Y per cent in the class five exam’ to saying ‘Student X has demonstrated the following levels of absolute competence in various topics/domains’. The levels can range from ‘not demonstrated’ to ‘mastered’ and would aim to provide feedback to parents and students (and eventually employers) about functional competencies as opposed to arbitrary marks.

In the long term, such assessments can be administered through technology-based platforms – which would enable dynamic adaptive tests and also ensure integrity of measurement through large item banks. But, in the short term, such assessments can be carried out through pre-printed worksheets for various topics.

This is an incredibly important reform to ensure that the millions of children who are behind grade level and at risk of being left behind permanently are encouraged to make ‘absolute’ progress regardless of their level. It is also a critical enabler of skilling and vocational education for students who may not take an academic track.

A key challenge for the skilling sector in India today is that the students who enter skilling or job-training programmes have very poor literacy and numeracy – and are thus often not equipped even to handle the curriculum of the skilling programmes. This is because by the time students are earmarked for vocational tracks in school and directed to such programmes, they have already fallen far behind the curriculum and have weak foundational skills.

Having credible signals of absolute credentials of learning will help students, parents, teachers, providers of higher education (including vocational and skill-based education), as well as employers.

3. A national mission to achieve universal functional literacy and numeracy: The single biggest failing of the Indian education system is the fact that seventy years after Independence, the majority of children completing primary education are still not functionally literate and numerate. This is both an economic and a moral failure. The lack of such foundational literacy and numeracy both inhibits the skill formation needed for economic growth and also robs millions of children and youth of the opportunity to participate in the broader economic growth of the country – and to become empowered citizens who are able to navigate a fast-changing world.

We believe therefore that the single most important outcome that education policy needs to deliver on for the future of the country is to ensure, by 2022, universal functional literacy and numeracy of all schoolchildren by the end of grade three. Indeed, the history of Indian education policy suggests that trying to do too much may have prevented it from achieving even the basics.

Thus, even if the entire education system is able to achieve just this one goal in the next few years in mission mode and establish processes to make sure that every cohort entering school starting in the academic year 2019–20 has achieved universal functional literacy and numeracy by the end of third grade, that will be an enormous success.

Achieving this goal will be enabled by three key sets of investments. The first is universal preschool education to ensure better school readiness by the start of first grade (described in further detail below). The second is providing supplemental instructional support to children who are falling behind in the early grades itself to ensure that every child is functionally literate and numerate by the end of grade three (this is especially important for the children currently in school). The third is investing in independent measurement and monitoring of the achievement of these goals (at least at the district level) and motivating the entire district education machinery to achieve these goals through a suitable combination of recognition and rewards.

The most essential immediate (and easily actionable) step to enable this is supplemental instructional support in the early grades. This support will be for small-group instruction that is pitched appropriately at the level of the student. Consultations with teachers clearly indicate that they are aware of this need – but are not able to simultaneously complete the curriculum in the textbook while also providing such supplemental instruction for children falling behind. At the same time, a large body of high-quality evidence (based on randomized controlled trials) from multiple states in India has shown that it is possible to achieve rapid progress in foundational literacy and numeracy within a few months provided children are taught in small groups at the right level – even if this instruction is delivered by modestly educated (secondary or higher secondary school level) youth from the same area (Banerjee, et al. 2007; Banerjee, et al. 2017).

We therefore recommend a budget be provided to every school for the hiring of part-time tutors (who could even be students in secondary or higher secondary school from the same village/area) to be able to provide such supplemental instruction. Such tutors can work under the guidance of regular teachers and provide one to two hours of supplemental instruction every day to students grouped by their learning level. Combined with independent monitoring of learning outcomes, such an approach is likely to be successful in a cost-effective way.

Note that we do not recommend the return to a para-teacher or ‘shiksha karmi’ model (which has weaknesses on multiple fronts including professional, legal and political). Rather, the extra budget to hire part-time tutors can be seen as a way of supporting a five-year national mission to ensure universal foundational literacy in numeracy.

In the medium to long term, as part of a systematic reform of teacher training and preparation, we recommend a new approach to pre-service teacher training that emphasizes substantial amounts of practical teaching as part of the training/credentialing process (see details below). Over time, we recommend that the time spent in such practical training by students training to become teachers can be focused on providing the small-group supplemental instruction needed to achieve this goal.

4. Universal preschool education to support school readiness before first grade: Large gaps in learning levels emerge even in the early years of schooling, attributable in large part to the substantial increase in first-generation learners in the schooling system. Thus we strongly recommend a national commitment towards universal preschool education to support school readiness and help achieve the goal of universal functional literacy and numeracy by grade three.

In practice, there are two different approaches to implementing this goal. The first is to add a year of kindergarten to the school education system. The second is to strengthen the quality of early childhood education in the anganwadi centres, by adding an extra anganwadi worker who is dedicated to delivering early childhood education (which will allow the current worker to continue focusing on child nutrition and health). There are advantages and disadvantages to both approaches, as outlined below.

The main advantages of the anganwadi-based model include:

a. Greater proximity to households (this is important because attendance is highly sensitive to distance at younger ages).

b. The curricular needs of early childhood education are substantially different from those of school education and need to focus more on play, self-regulation, social skills and school readiness more generally. This may require specific forms of training and may be better delivered in an anganwadi centre. There is a risk that preschool education that is based in schools may simply result in standard classroom instruction being conducted at a younger age (since such a structure is likely to draw on teachers who are currently teaching primary grades).

c. Since anganwadi workers are typically hired locally in the same village, they are both more connected to the local communities and likely to have lower rates of absenteeism (compared to teachers who typically live in urban areas and commute to rural postings).

d. Finally, the anganwadi-worker-based model is likely to be much more cost-effective (even after adding an extra anganwadi worker to each anganwadi centre to focus on early childhood education) than a model based on hiring more regular teachers in government schools. Given the fiscal constraints in expanding early childhood education, this is a very important consideration and is likely to make it easier to scale across the country.

i. For instance, based on recent research in Tamil Nadu on the impact of adding an extra worker to anganwadis to focus on early childhood education, we estimate that the present discounted value of the policy was Rs 16,000–20,000 per month. Thus, the investment in the extra worker would be cost-effective at a monthly salary under this range but not above.

ii. At present costs, it would be cost-effective to add an anganwadi worker to focus on early childhood education (since average monthly anganwadi worker salaries range from Rs 4000–10,000). But given regular teacher salaries of Rs 30,000–60,000, the school-based model would not be cost-effective.

However, the primary-school-based model too has some advantages. These include:

e. Greater ease of attendance for siblings when the older child is of schoolgoing age, while the younger child is of preschool age.

f. Greater scale in terms of facilities.

g. Potentially doable at lower cost if the system has a large number of ‘surplus’ teachers due to declining enrolment in government schools (but this should likely not be a reason to expand regular teacher hiring).

h. Finally, the quality of the anganwadi system varies widely across states and it is possible that the anganwadi system in some states is too dysfunctional to rely on for early childhood education.

Thus, while the factors above point towards the likely superiority of the anganwadi-based model, it is important to recognize the variation across states and not impose a specific model – as long as states deliver on universal early childhood education.

5. Reform of teacher training to emphasize pedagogy over theory and the inclusion of extensive practical training. Pre-service training: The teacher is the single most important determinant of the quality of education received by the student, and it is hoped that having qualified/credentialed teachers will help improve the quality of education – indeed, a cornerstone of the RTE is a requirement that all teachers possess a formal teacher training credential.

Unfortunately, this requirement is not supported by the evidence. In particular, several studies have shown that there is no correlation between a teacher possessing a formal teacher training credential and his or her effectiveness in the classroom as measured by improvements in learning outcomes during the period that students are with that teacher (Kingdon and Teal 2010; Muralidharan and Sundararaman 2011; Muralidharan 2012).

Of course, this does not imply that teacher training cannot be effective. However, the evidence is also very clear that the status quo of teacher training is broken and ineffective (this point is also made clearly in the Justice Verma Commission report of 2012). Two of the several reasons for this are particularly important.

The first is the poor quality of the majority of teacher training institutes, especially the ones based on distance learning or correspondence courses (which have no practical training at all). Second, a content analysis of even the better teacher training programmes indicates that the curriculum mostly emphasizes the history, theory, sociology and philosophy of education and has very little focus on pedagogy and also very little practical training. In contrast, global evidence suggests that the most effective forms of training in professional roles involves extensive practical training and learning on the job (Muralidharan 2016).

We therefore recommend a new paradigm for teacher training and professional development that places much more emphasis on practical training through apprenticeship as part of the teacher credentialing process. Specifically, we recommend that the ideal structure for pre-service teacher training should intersperse modules of theory with extensive amounts of practicum-based training that will lead to formal teacher training credentials that reflect this practical experience and learning.

We recommend the setting up of a few apex teacher training institutes in each state (in highly reputable public and non-profit private institutes) to design such a programme. We also recommend that such practicum-based teaching degrees are four years long and admit students after class twelve. For elementary school teachers this would result in a B.El.Ed. (practice) degree and feature at least six months of classroom-based practical training each year so that over a period of four years at least half the time is spent on training in the classroom. For secondary-grade teachers, such a programme could be five years long and combine a three-year bachelor’s degree in a subject with a year of education theory and a year of practical training.

We also recommend a close integration of such training programmes with the public schooling system, so that trainees are effectively integrated into the schools where practical training will take place and can provide meaningful instructional support in these schools. Consistent with RTE norms, the trainees will not be responsible for classes on their own, but will function under the close guidance and supervision of the regular teachers and assist in instructional tasks (especially small-group instruction to support the achievement of universal functional literacy and numeracy for all primary school students).

During the period of practical training, we expect that the training institution will continue to engage with the students through online and smartphone-based interactions – including viewing videos and answering quizzes, writing reflections of their teaching experience and participating in virtual communities of practice with other trainees. Thus, the goal is to achieve a deep integration between theory and practice and reflection on how the two inform each other during the course of pre-service teacher training.

We expect such a teacher training programme to be both prestigious and coveted since (a) the number of places will be limited and admission will be restricted to top scoring twelfth standard graduates, (b) the fees will be waived/subsidized for many students because the practical training includes an element of serving underserved areas, and (c) students will also get a modest stipend (paid for by the education department) for their months of practical training when they will be working in schools.

To address the challenge of spatial mismatch between where the academically strongest students are (typically urban areas) and where the teaching needs in the government schools are the greatest (typically rural areas), we recommend that admission to such a prestigious integrated teacher training programme be geographically dispersed.

For instance, one approach would be to admit the highest scoring applicant from each panchayat into the programme, with the understanding that the practical training will be based in a government school in the same panchayat. A further advantage of such an approach is that it enables greater female participation in rural areas by providing training/employment opportunities in the same village. Various aspects of reservations and quotas can also be implemented but may need to be achieved at the district or block levels.

Over time, we also recommend that the process of hiring regular teachers give extra points for each year of actual teaching experience (with years spent as part of practical training counting for credit). Our vision therefore is that in the long term almost all teachers hired into the regular government teaching positions will have completed such integrated practicum-based teacher training programmes.

Such an approach will have several advantages including:

a. Improving the quality of pre-service teacher training by requiring actual pedagogical experience as part of the credentialing process.

b. Improving the ability of trainee teachers to absorb the significance of the theoretical content that is present in typical programmes by observing how these issues matter in practice.

c. Ensuring that teachers entering the education system with permanent jobs (as is typical for government teachers) would have had adequate experience in classroom management and functioning as teachers before obtaining lifetime appointments.

d. Providing a cost-effective way of augmenting instructional resources in government schools. In particular, this would be a more sustainable way of enabling schools to have the teaching resources to provide small-group supplemental instruction for first-generation learners to ensure universal foundational literacy and numeracy for all future cohorts of children in the Indian education system.

We believe that implementing such an approach to pre-service teacher training can be transformative in improving the long-term quality of Indian education both through its impact on improving the pipeline of entrants into the teaching profession and through its immediate impact on ensuring universal foundational literacy and numeracy (through the supplemental instruction provided by the trainees as part of their practical training).

While this approach can help in improving the long-term quality of teachers, urgent attention is also needed to be given to the problem of in-service training for the hundreds of thousands of incumbent teachers, many of whom have entered the system with very poor pre-service training. The biggest challenge in implementing such in-service training is the complete lack of visibility on the quality of such training. Thus, while most states’ teacher policies provide for around twenty days a year of in-service training, in practice these trainings are of extremely variable quality and our interactions with stakeholders suggest that these are typically not effective.

Thus, improving the quality of in-service training needs to focus on a few key principles. First, emphasizing effective pedagogy (especially through sample demonstration lessons to illustrate various aspects of effective pedagogy including content, engagement, inclusion and student support). Second, it needs to curate the quality of the in-service training programmes. Third, it needs to ensure that the training is actually done as per the regulatory norms and that teachers are assessed on their comprehension of the training content.

We recommend that the most promising practical way of achieving these goals is to make much more use of online teacher training content. Specifically, we recommend developing a portal for in-service teacher training that can host thousands of videos (including translations in all major Indian languages) and training modules for various topics that are relevant for in-service teacher training. For instance, a training module could illustrate how to teach the class in an interactive way that engages students, as opposed to simply reading from the textbook and writing on the blackboard, and demonstrate this with short videos. Others could illustrate effective ways of teaching specific concepts. Yet others could illustrate inclusive teaching practices that engage all students and not just the academically stronger ones.

Such a portal would be able to achieve all the principles outlined above. The modules would emphasize pedagogy and provide the training in modular bits as opposed to day-long programmes (research strongly suggests that students are more likely to absorb new material in capsules rather than traditional lecture-based instruction). The portal would allow teachers to rate the quality and usefulness of different kinds of content and therefore create a crowdsourced way of identifying high-quality content (which would reflect the wisdom of teachers around the country). Finally, by providing each teacher with a unique ID to access the portal, it becomes easy to track usage, absorption of materials (through short quizzes at the end of modules) and completion of annual in-service training targets. Over time, modules can be organized into courses, and demonstrated mastery of the content of in-service training courses can become an input into identifying effective teachers and promoting them into roles of greater leadership and responsibility.

The Diksha portal developed by the Ministry of Human Resource Development implements these principles for content creation, but delivering full potential will also depend on integration into teacher evaluations, increments and promotions along the lines outlined above.

6. Private school regulation and RTE reforms: While Central and state departments of government mostly focus on government-run schools (for which they have implementation responsibility), it is imperative that education policy accounts for the large prevalence of private school providers. Recent estimates suggest that over 40 per cent of school enrolment in India is in private schools, with the share being over 70 per cent in several large cities (FICCI and Ernst & Young Report 2014). The total share of private school enrolment in the twenty largest states is around 55 per cent at the secondary and higher secondary levels.

A national education policy should focus on the quality of education received by every student in India and not distinguish between a student enrolled in a public or in a private school. Thus, a critical enabler of improving education quality in India is expanding the supply of high-quality institutions (both public and private).

However, there are severe constraints on the entry of high-quality private school providers – most notably placed by the RTE. We now have data to assess the implementation of the RTE to offer an informed view on how this has affected education quality and availability in India.

There are two important reforms to RTE provisions on private schools that we recommend:

a. Regulation of private schools based on disclosure and not based on input mandates: The RTE’s input-based approach to education quality never made sense given the extensive evidence that most school inputs are neither necessary nor sufficient for improving learning outcomes. This has led to an unnecessary and disruptive closure of several low-cost private schools that parents were choosing of their own accord. In many cases, even government schools are in violation of these input-based norms. We therefore recommend:

Repealing all input-based mandates for schools under the RTE (for both public and private schools) and changing the approach to regulation of private schools to be based on transparency and disclosure as opposed to input-based mandates. By focusing regulation on disclosure, policy would acknowledge the considerable variation across India and allow diversity of models of effective schooling to emerge. Regulation is still important, and private schools can and should still be sanctioned for lying, but they should not have to meet input mandates.

Such an approach will facilitate (as opposed to inhibit) the expansion of quality private-school providers and allow for variation in approaches across locations and providers. As an aside, it would also facilitate localized cost-effective innovations by government schools, which may be made difficult by the RTE (such as hiring tutors without formal teaching credentials for providing supplemental instructional support).

b. Introduce a national policy for charter schools (that will at least permit serious pilots): The goal of RTE Clause 12c was to provide students from economically weaker sections (EWS) the opportunity to attend private schools, and it aimed to reimburse private schools for this.

The fundamental problem with this approach is that it was based on confiscating existing capacity in private schools for a social purpose as opposed to creating new capacity. Further, it was deeply illiberal in that it was basically a stealth nationalization of a quarter of private schooling capacity without adequate compensation. Finally, data over the past five years show that several states have (a) set the reimbursement rates considerably below the norms in the law, and (b) in many cases not reimbursed private schools in a timely manner.

The consequence has been a steady shutting down of non-minority private schools that have not been able to bear the burden of the cross-subsidy imposed on them by the RTE.

We therefore propose that RTE Clause 12c be replaced with an approach to public–private partnerships that achieves the equity aims of RTE Clause 12c while also increasing the supply of high-quality education options in India.

Specifically, global evidence suggests that charter schools – which are public schools that do not charge fees (because they receive public funding equal to the per-child variable cost in the public system) and cannot admit students selectively, but managed by private entities with operational autonomy (especially over teacher hiring and accountability) – have been successful at substantially improving outcomes for disadvantaged students while also increasing the total supply of schools.

Indian evidence (Muralidharan and Sundararaman 2015) suggests that private schools are more productive than public schools (delivering similar or modest improvements in learning at a much lower cost per child). Thus, if privately managed schools (with autonomy over teacher hiring, retention, pay and performance management) were to have the same level of per-child spending as the current government schooling system, we could potentially substantially improve learning outcomes without increasing the spending per child from the status quo (although there is no direct evidence of this to date). There are hundreds of high-quality operators of private schools who would be willing to operate hundreds or thousands of new schools (or take on management contracts of existing government schools) that serve EWS students if offered a reliable per-child compensation by the government that is of similar magnitude to the current spending per child in government schools.

Of course, leveraging such operators for providing a public interest good like education will require adequate oversight and regulation. But an enabling framework for such an approach could put India on a qualitatively different path for improving education outcomes in the coming decade. At the very least, policy guidelines to this effect can motivate serious pilots and evaluations of this approach.

7. Taking vocational education seriously: The jobs crisis in India is partly a skills crisis with millions of ‘educated’ unemployed youth on the one hand and employers routinely complaining that they cannot find adequately skilled manpower. A major reason for this is the focus of the education system on passing exams (usually by cramming) with no real understanding of the subject matter.

While there is not enough research on vocational education in India to have evidence-backed ideas for policy, there is suggestive evidence from the United States of the importance of integrating vocational education into secondary and higher secondary school curricula. Recent research suggests that the rising ‘college wage premium’ (defined as the wage increase of a college graduate relative to a high school one) may not be being driven only by a more complex economy and the greater need for higher education (as is commonly believed). Rather, it suggests that this may be explained at least in part by changes in high school curricula. Specifically, Alon (2018) shows that high school curricula in the United States used to have substantial vocational content up to the 1950s but that this changed in the 1960s to focus almost exclusively on college preparation.

While this made sense for students who did go to college, it may have worsened labour market outcomes for those who did not – since their high school education did not suitably prepare them for the labour market (compared to previous curricula that had more vocational content).

It is important for India not to repeat this mistake. Vocational education is widely seen as not desirable and something that is only chosen by students who are not academically smart. This will have to change if we are to provide the human capital needed to enable inclusive growth.

Some of the leading education systems in the world, including those in Singapore, Germany and Switzerland, feature tracking of students into vocational streams after grades six or seven. This allows these students to obtain more human capital than they would obtain by staying in school, through a combination of vocational classroom training and practical training or apprenticeships. The training is also linked to credible credentialing (which is sorely missing in our current skills ecosystem). This enables markets to pay a wage premium for skills, and qualified workers in technical fields can earn middle-class wages and incomes.

India would do well to think seriously about such an approach, which would likely better serve the millions of students who ‘pass’ exams but cannot find jobs because they have no real skills.

Appendix A: Heterogeneity and Learning Levels in Middle and Primary Grades

The figures on pages 161 and 163 show the levels and dispersion of student achievement in mathematics and Hindi in samples of students from two states, Delhi and Rajasthan. This comes from two separate studies using the Mindspark software (developed by Educational Initiatives, an Ahmedabad-based company) with government school students in these states.

The graphs rely on two pieces of information: (a) the grade that students are enrolled in and (b) the software’s assessment, based on a common diagnostic test given to all students at the beginning of the intervention which assesses their actual ability levels. This can be thought of as the software’s assessment (based on all questions answered) of the achievement level of the child (and is the level at which the software will begin to pitch instruction afterwards).

Figures 1 and 2 come from a sample of students from five government schools in Delhi who had chosen to take part in an evaluation study of Mindspark and had been selected randomly by lottery (and are reproduced from Muralidharan, Singh, and Ganimian 2017).

Note that the main reason these data rely on computer-based assessments is that the tests are administered without ceiling or floor effects. This dynamic testing allows us to measure the exact learning level of each student. In contrast, paper-and-pencil tests (typically of grade-level content) suffer from severe floor effects in the Indian context. Thus, if a student has a very low score on such a test, we have no idea as to how far below grade-appropriate competence the student is (whereas this is not a problem on a dynamic computer-based test).

Figures 1 and 2: Learning levels in Delhi schools

There are three key patterns illustrated by these figures:

1. By the beginning of grade 6, students are, on average, 2.5 years behind in maths.

2. By grade 9, this gap is even larger, at about 4.5 years in maths and 2.5 in Hindi.

3. In any given grade, the learning levels of students span four or five grade levels.

The patterns shown in these figures may well be the most important facts about education in India. While the facts about low average levels of learning are well known, these figures illustrate two additional critical points (2 and 3). In particular, the striking variation in student learning levels within a class highlights both how far behind grade-level curricular standards many students are and also the extremely challenging task for teachers in government schools in handling such variation in student achievements.

Consistent with the patterns in these figures, Muralidharan, et al. (2019) also find that students in the lowest third of within-grade test scores make no progress at all in learning during the school year, despite being enrolled in and attending school (consistent with their being so far behind curricular standards that textbook-based classroom instruction is essentially useless).

These facts are implied by several other studies across India as well, and are likely to be an important reason for why the very large increases in education expenditure and resources in the past two decades have not translated into improved learning (because these resources do not address the binding constraint of the education system, which is that children are too far behind).

One limitation of Figures 1 and 2 is that the sample is limited, coming only from five schools in one city and, further, from a non-representative set of students who chose to be part of the study.

Figures 3 and 4 draw instead on a much larger sample of over 5000 students in forty Adarsh schools in four districts in Rajasthan, covering both urban and rural areas and spanning the entire range of elementary school from classes one to eight (from an ongoing study of the impacts of deploying the Mindspark software in government schools in Rajasthan).

Figures 3 and 4: Learning levels in Rajasthan Adarsh schools

All students in these classes in the programme schools were covered (and hence we do not need to worry about self-selection of better/worse performing students here). As we can see, the entire pattern observed in the smaller sample in Delhi is evident in Rajasthan too: there is similar dispersion in classes six to eight as in Delhi, and similar deficits compared to curricula which widen over this period.⁷

Much more importantly, we see that this process of widening spans the entire period of elementary school with the deficit from grade-appropriate standards and the dispersion within a classroom both increasing sharply with every additional year.

These graphs illustrate the most critical constraints in the structure of the Indian education system today: the curriculum targets only the very top of the distribution and leaves most students behind; the immense variation within a classroom makes the delivery of any effective instruction very hard; and, consequently, most students are far from class-appropriate standards even after completing the full course of elementary education.

References

T. Alon, 2018, ‘Earning More by Doing Less: Human Capital Specialization and the College Wage Premium’, Northwestern University.

A. Banerjee, S. Cole, E. Duflo, and L. Linden, 2007, ‘Remedying Education: Evidence from Two Randomized Experiments in India’, Quarterly Journal of Economics, 122: 1235–64.

A. Banerjee., R. Banerji, J. Berry, E. Duflo, H. Kannan, S. Mukerji, M. Shotland, and M. Walton, 2017, ‘From Proof of Concept to Scalable Policies: Challenges and Solutions, with an Application’, Journal of Economic Perspectives, 31: 73–102.

FICCI and Ernst & Young, 2014, ‘Private Sector’s Contribution to K-12 Education in India: Current Impact, Challenges, and Way Forward’.

G. Kingdon and F. Teal, 2010, ‘Teacher Unions, Teacher Pay and Student Performance in India: A Pupil Fixed Effects Approach’, Journal of Development Economics, 91: 278–88.

K. Muralidharan, 2012, ‘Long Term Effects of Teacher Performance Pay: Experimental Evidence from India’, University of California, San Diego.

——, 2013, ‘Priorities for Primary Education Policy in India’s 12th Five-Year Plan’, India Policy Forum, 9: 1–46.

——, 2016, ‘A New Approach to Public Sector Hiring in India for Improved Service Delivery’, India Policy Forum.

K. Muralidharan, A. Singh, and A. Ganimian, 2019, ‘Disrupting Education? Experimental Evidence on Technology-Aided Instruction in India’, American Economic Review.

K. Muralidharan and V. Sundararaman, 2011, ‘Teacher Performance Pay: Experimental Evidence from India’, Journal of Political Economy, 119: 39–77.

——, 2015, ‘The Aggregate Effects of School Choice: Evidence from a Two-Stage Experiment’, Quarterly Journal of Economics, 130: 1011–66.

Joppe de Ree, Karthik Muralidharan, Menno Pradhan, and Halsey Rogers, 2018, ‘Double for Nothing? Experimental Evidence on an Unconditional Teacher Salary Increase in Indonesia’, The Quarterly Journal of Economics, 133(2): 993–1039.

The Solutions

1. A national mission to ensure, by 2022, that every child can actually read, write and do simple maths by the time they have finished class three. For this, hire part-time primary school teachers – trainees or even high-school students – to focus on those being left behind.

2. Introduce universal preschool, perhaps through anganwadi childcare centres. This will ensure that children from families receiving formal education for the first time do not start at a disadvantage that grows with time.

3. Change how teachers are trained, by increasing the emphasis on practical, hands-on experience.

4. Reform the RTE so that private schools can be supported and focus on the educational outcomes they deliver rather than on a rigid set of requirements for schools. Also change high school education to introduce a strong vocational education and apprenticeship track.

*These are the author’s personal views and reflect inputs provided to the committee writing the New Education Policy for India and to the ‘Economic Strategy for India (2018)’ put together by a non-partisan group of economists. This version is as of 15 February 2019.

¹As shown in Muralidharan, Singh, and Ganimian (2019).

²This is why well-intentioned ideas such as scrapping the tenth board exam are highly counterproductive in practice. Scrapping the signal does not remove the need for one and it ends up disproportionately hurting disadvantaged students who do not have other options for signalling their talent.

³A particularly striking example is provided from a high-quality study of unconditional teacher salary increases in Indonesia, where the government doubled teacher pay in 2005 – but this very large increase led to no improvement in student learning (de Ree, et al. 2017).

⁴For instance, exams for selecting bureaucrats in ancient China required the memorizing of extensive amounts of Confucian text. Mastery of these texts would have likely provided only limited training in ‘how’ to be a better administrator, but the sheer volume of content ensured that only students of naturally higher ability were likely to pass the exam. This is a classic example of a sorting-based system as opposed to a human development one.

⁵For secondary and higher secondary school, a related concept is to try to follow the US system a bit in terms of ‘modularizing’ the content of maths and science syllabi in high school. A certain amount of tracking is then made feasible, with students working their way through ‘modules’ at a variable pace as opposed to the entire class following the same textbook and syllabus with no differentiation.

⁶A great example of this is provided from a recent study in Delhi of the effectiveness of an after-school computer-aided learning programme (Muralidharan, Singh, and Ganimian 2019). The programme was incredibly effective in improving learning outcomes of upper primary students in grades six to nine targeted by the programme (way more effective in a faster time than any other intervention evaluated with comparable rigour in India). However, these learning improvements were taking place at a level of learning that was two or three grade levels below the currently enrolled grade. Thus, the substantial improvements in learning were not seen on the grade-appropriate school exam. This may explain why parental demand for the programme was low even though it was so effective at improving learning.

⁷The one change is that there are now no students assessed at above grade level (unlike in Hindi in Delhi). This is an artefact of the algorithm which was modified in 2017 to cap assessed student achievement at their current grade (since the primary purpose of this metric is to fix the level at which instruction will be presented to students).

11