Enrolled and Unable to Read

Between September and November 2024, volunteers from local organisations in 605 rural districts knocked on doors in 17,997 villages and asked to see the children. They reached 649,491 of them. In each sampled household, every child aged five to sixteen was given the same two tasks, one at a time, sitting down: read as far as you can down a card that runs from letters to words to a short paragraph to a story, and do as much as you can of a set that runs from recognising the numeral 7 to dividing a three-digit number by a single digit. The story on the card is written at the level expected of a child finishing Std II. The tests have not changed since 2006, which is the entire point of them.

Two of the findings sit next to each other in the report. Enrolment among children aged six to fourteen was 98.1 per cent, and has been above 95 per cent for close to twenty years. Of the children enrolled in Std V in government schools, 44.8 per cent could read the Std II story.¹

Neither number is disputed. They are produced by the same enumerator, in the same visit, about the same child.

Where the survey stands

The Annual Status of Education Report is not an official statistic. It is run by Pratham, a non-governmental organisation, through local partners, and it has no legal standing at all. Its design choices follow from where it stands, and they are worth setting out because they explain why it finds what the state's own systems do not.

The survey is conducted in households rather than schools. A school-based assessment tests the children who are present in the classroom on the day the assessors arrive, which excludes the ones who have dropped out, the ones enrolled but not attending, and the ones absent that morning. Average student attendance in government primary schools was 75.9 per cent in 2024, so a school-based test in that year missed roughly a quarter of the enrolled children by construction, and the quarter it missed is not a random draw.¹

The test is set deliberately low. Asking whether a Std V child can read a Std II text is not an attempt to measure achievement against the curriculum; it is an attempt to find the floor. A curriculum-referenced test in Std V would return a distribution of scores that could be discussed. A floor test returns a proportion who cannot do the thing, and proportions are harder to argue with.

The state's own instruments are built the other way round. The Unified District Information System for Education records enrolment, teachers, classrooms, toilets and books, collected from schools by the people who run them. The National Achievement Survey tests learning, but does so in schools, in a sample of grades, once every several years. Between them they measure the inputs continuously and the outputs occasionally, and the continuous measurement is the one that gets used.

The enrolment figure also conceals a good deal of movement. In 2018, 65.6 per cent of children aged six to fourteen were in government schools. During the pandemic that rose to 72.9 per cent, as households that had been paying private fees stopped being able to. By 2024 it had fallen back to 66.8 per cent, in every state except Uttarakhand and Jammu and Kashmir.¹ Roughly seven per cent of rural children changed sector twice in six years, and the headline enrolment rate moved by a little over one percentage point across the whole episode. A number that stable is not measuring nothing. It is measuring one thing, attendance on a roll, and it is doing so accurately.

Households, meanwhile, have been buying what the entitlement does not deliver. The proportion of children in Std I to VIII taking paid private tuition rose from 26.4 per cent in 2018 to 30.5 per cent in 2022, with increases of eight percentage points or more in Uttar Pradesh, Bihar and Jharkhand.¹ Free and compulsory education, in the sense the Act guarantees it, is being supplemented out of pocket by close to a third of rural families, most of them poor.

What the law made enforceable

That division was written into the statute.

The Constitution (Eighty-sixth Amendment) Act, 2002 inserted Article 21A, making free and compulsory education for children aged six to fourteen a fundamental right. The enabling legislation followed seven years later: the Right of Children to Free and Compulsory Education Act, 2009, passed by the Lok Sabha on 4 August 2009, assented on 26 August, and brought into force on 1 April 2010.

The Act's Schedule sets out what a school must have. Pupil-teacher ratios by enrolment band. A classroom for every teacher. An office for the head teacher. A barrier-free ramp, separate toilets for boys and girls, drinking water, a playground, a library, teaching-learning material, minimum working days and minimum instructional hours per year for each stage. Section 16, as originally enacted, prohibited holding any child back or expelling them before completing elementary education. Section 24 prescribes teacher duties. Section 25 sets the ratio.

There is no provision that says a child leaving Std V shall be able to read.

This is not an oversight, and it is defensible on ordinary legal grounds. An enforceable standard needs a duty that is observable, attributable to a duty-bearer, and capable of being remedied by an order of a court. A pupil-teacher ratio of thirty to one satisfies all three. A reading level does not: it is jointly produced by the school, the household, the language of instruction, the child's health and nutrition, and the years that came before, and a writ court cannot order any of these into existence. The drafters chose the duties that could be litigated.

The consequence is that India's education system has a legal obligation to seat children and no legal obligation to teach them anything in particular. It also has a financing system keyed to the same variables. Grants flow against enrolment, infrastructure and teacher posts, all of which are counted every year with reasonable accuracy. What is measured well is what is owed, and what is owed is what is funded.

There is a further problem the Schedule did not anticipate, and ASER's school visits pick it up. The proportion of government primary schools with fewer than sixty children enrolled rose from 44 per cent in 2022 to 52.1 per cent in 2024, and in Jammu and Kashmir, Himachal Pradesh, Uttarakhand, Nagaland and Karnataka more than eighty per cent of primary schools are now that small. Two-thirds of Std I and Std II classrooms had children from more than one grade sitting in them.¹ The Act's norms are specified per school and per teacher. As enrolment thins out across a fixed stock of buildings, a school can satisfy every input norm in the Schedule and still put two grades in one room with one teacher, which is a teaching problem the ratio does not describe.

The gradient

The second chart shows the same reading test given to three different grades in government schools. Read it downwards rather than across.

In 2024, 23.4 per cent of Std III children could read the Std II story. By Std V the figure was 44.8 per cent, by Std VIII 67.5 per cent.¹ Children do improve as they move up. But the test is fixed at the Std II level throughout, so the Std VIII figure means that roughly a third of children in their eighth year of school could not read a text written for their second. The 2024 Std III result is the highest recorded for government schools since ASER began in 2005, and it still leaves more than three-quarters of the grade below the line.

The gap between government and private schools narrowed over the same period, and not in the direction anyone wanted. Std V reading in private schools fell from 65.1 per cent in 2018 to 56.8 per cent in 2022 and recovered only to 59.3 per cent in 2024, while the government school figure returned to its 2018 level.¹ Private schooling, which absorbs about a third of rural children and is bought at a price, delivered a worse recovery than the sector it is chosen in preference to.

Arithmetic follows the same shape and is worse. Nationally in 2024, 33.7 per cent of Std III children could do a two-digit subtraction with borrowing, 30.7 per cent of Std V children could do a three-digit-by-one-digit division, and 45.8 per cent of Std VIII children could do that division.¹ The Std VIII arithmetic figure has moved by less than two percentage points across three surveys spanning six years and a pandemic.

The 2022 round measured what the school closures did. Std III reading in government schools fell to 16.3 per cent, taking national reading levels back below where they had been in 2012. Std V fell to 38.5 per cent. The 2024 round found most of that recovered and, in Std III, more than recovered. Attributing the recovery is harder than measuring it: the foundational literacy mission announced in 2021 was running through this period, and so was a large expansion of pre-primary enrolment, with the proportion of five-year-olds in some pre-primary institution rising from 62.1 per cent in 2022 to 71.4 per cent in 2024.¹

What the no-detention debate was actually about

Section 16 became the visible fight, and it is a good illustration of how a measurement problem gets argued as a discipline problem.

The provision was defended on the ground that holding children back pushes them out of school, which the evidence supports, and attacked on the ground that automatic promotion removed any check on whether learning had occurred. Both are true, and they are true because the system had no other check. If the state had measured reading in Std III every year and acted on it, the examination in Std V would not have been asked to carry the entire burden of quality assurance.

Parliament amended the Act in 2019 to permit an annual examination in Classes V and VIII, with remedial instruction and a re-examination before any child could be held back. The amendment restored a filter. It did not create a measurement.

The asymmetry

The position, then, is this. The state counts, with considerable care and at considerable expense, the number of children sitting in classrooms, the number of teachers in front of them, and the number of toilets behind the building. It has a statutory duty to provide those things and it is held to that duty. The question of whether the children can read is answered by a survey run by a charity, using a test devised in 2005 and deliberately kept unchanged, funded outside the government, with no power to compel anything at all.

That is not a failure of the survey. ASER does its job, which is why its numbers are quoted in Parliament and in the Economic Survey. It is a fact about which measurements the state built into its own machinery and which it left to others. The 2024 report shows enrolment at 98.1 per cent, and it shows 23.4 per cent of Std III children in government schools reading at the level of the grade below the one they finished two years ago. Both figures were collected in the same room.

1. ASER Centre, Annual Status of Education Report (Rural) 2024, national findings, released January 2025. Enrolment, reading, arithmetic, attendance and pre-primary figures are from that document. Figures for 2022 are from ASER 2022 national findings. ↩