Vella_Theory_Indian_Education_Research

VELLA THEORY · RESEARCH BRIEF

Why Are Indian Children Not Learning?

Enrollment Without Learning — Inside India's K-10 Classrooms

Prepared: September 2026 · 59 sources · Every claim linked inline · 3-part investigative structure

Research only — no narrative script. Story framing at author's discretion.

CONTENTS

TL;DR

Key Findings

PART 1 — The Visible Problem

1.1 The Observation — Enrollment Is Universal, Learning Isn't

1.2 Suspect One: Not Enough Teachers?

1.3 Following the Crack — Absenteeism, Non-Teaching Duty, Deployment

1.4 The Infrastructure Question

1.5 Who's Actually Teaching? — Recruitment Integrity

1.6 The Government's Own Policy Swings

PART 2 — The Twist: What the System Cannot Fix Alone

2.1 The Question Nobody Puts in the Report

2.2 Before the School Gate — Early Childhood Readiness

2.3 The Home Half of Learning

2.4 Enrolled ≠ Engaged — The Attendance Gap

2.5 The Authority Question — Discipline and Corporal Punishment

2.6 When the Classroom Loses Control

PART 3 — What Actually Works

3.1 Teaching at the Right Level (TaRL)

3.2 The Post-Pandemic Surge

3.3 What This Means for the System as a Whole

Recommendations for the Documentary

Caveats

Sources & References

TL;DR

The core answer: Indian children reach Class 10 without learning because the school system optimised for the countable inputs it could easily deliver — enrollment (~98% for ages 6-14) [2] [1], meals, buildings (electricity 53%→92% over 2013-24) [18], and teacher headcount (now 1 crore, PTR within the RTE 30:1 norm) [1] — while the uncountable thing that actually produces learning, instruction aligned to each child's level, was left largely unaddressed. Government Class 5 reading at Class 2 level actually fell from 56.2% (2008) to 44.8% (2024) as infrastructure roughly doubled. [3] [2]

Institutional failures are real but insufficient: teacher absenteeism (~25% in 2003, ~23.6% a decade later; $1.5bn/yr fiscal cost) [11] [12], diversion to election/census duty [15] [16], 1.04 lakh single-teacher schools [17], and two decades of documented fake-degree recruitment fraud in Bihar (15,000 sacked in 2008; 3,000+ dismissed in 2026) and degree-mills like OPJS University (43,000+ forged degrees) [22] [24] [26]. But similar-resourced states differ by 40+ points in reading, proving resources aren't destiny. [36]

What demonstrably works is Teaching at the Right Level (large effect sizes across six Indian RCTs) [53] [54] and targeted remediation (which drove two-thirds of post-COVID recovery in six months and powered Uttar Pradesh's rise from 6% to 27.9% Class 3 reading, 2014-2024) [57] [37] — attacking the curriculum-pace/child-level mismatch rather than buying more inputs.

Key Findings

→ Enrollment ≠ learning is the central, measurable fact, and the gap is worsening on the longest time series (ASER Class 5 govt reading 56.2%→44.8%, 2008-2024). [3]

→ The teacher-shortage explanation collapses on inspection: national PTR now meets the RTE norm, but the workforce is maldistributed (Bihar 42:1 vs. Kerala/HP surpluses). [9] [1]

→ Infrastructure and mid-day meals improved access without improving learning — the strongest single refutation of "resources cause learning." [19]

→ Recruitment integrity is a documented, decades-long failure in specific states (Bihar fake-degree cohort 2006-2026; OPJS University 2024), with accountability spread across recruiters, degree-mills, regulators, and private managements. [21] [26]

→ The decisive variance sits outside the school — early readiness, first-generation-learner home gaps, the enrollment-attendance gap (~76% attendance vs. ~98% enrollment). [43] [40]

→ Two Part-2 debates are genuinely unresolved: whether the corporal-punishment ban (without alternative training) hurt classroom order, and whether "students filming teachers" has eroded authority at scale (evidence is anecdotal). [49] [52]

→ Targeted instruction is the evidence-based lever — but scale-up fails when bolted onto bureaucracy and succeeds only when the system reorganises around the child's level with real monitoring. [53]

PART 1 — The Visible Problem

1.1 The Observation — Enrollment Is Universal, Learning Isn't

India has, by any reasonable measure, won the battle for access to school and lost the battle for learning inside it. UDISE+ 2024-25, released by the Ministry of Education on 28 August 2025, records roughly 24.69 crore students enrolled across about 14.7 lakh schools, with the teaching workforce crossing 1 crore for the first time [1]. ASER 2024 — Pratham's household survey of 649,491 children across 17,997 villages in 605 districts — confirms enrollment for the 6-14 age group hovering near 98%. [2]

But when the same survey sits a child down and asks them to read, the picture inverts. In 2024, only 23.4% of Class 3 government-school children could read a Class 2-level text — celebrated as the highest figure since ASER began in 2005. [2] Among Class 8 students nationally, 45.8% could solve a division problem, barely moved from 44.7% (2022) and 44.1% (2018). [4]

The Long-Run Number That Reframes Everything

Government-school Class 5 students able to read a Class 2-level text fell from 56.2% in 2008 to 44.8% in 2024 — even as school infrastructure roughly doubled over the same period. [3] Because ASER's testing tools have been unchanged since 2006, this is a genuinely comparable decline, not a measurement artefact.

This is 'schooling ain't learning': a child can pass through a decade of attendance while acquiring skills the curriculum assumed by age eight. [2]

Complication: the problem is not that nothing improved. Class 3 arithmetic (subtraction) rose from 28.1% (2018) to 33.7% (2024) [5], and 2024 marked a real post-pandemic recovery. But recovery to a low baseline is not a system that teaches most children to read.

1.2 Suspect One: Not Enough Teachers?

The intuitive first suspect is teacher shortage — and the vacancy data appears to confirm it. A Parliamentary Standing Committee report in March 2023 flagged more than 9.8 lakh vacant teaching posts, over 7.4 lakh at the elementary level. [6] By December 2023 the Ministry's own figure was 8.4 lakh. [7] Secondary and higher-secondary vacancies nearly doubled in a single year, from 2.34 lakh to over 4.09 lakh — exactly the level where subject specialists matter most. [8]

Complication — the twist that reframes the suspect: nationally, the pupil-teacher ratio now comfortably meets the RTE norm at every level, and the teacher count crossed 1 crore, a 6.7% rise from 2022-23. [1] [10] So the aggregate teacher shortage cannot be the primary explanation for flat learning — if it were, learning would improve as PTR fell toward the norm, which it largely has not. The shortage is real but maldistributed: a national average of 21:1 at secondary coexists with Bihar's 42:1 at primary.

1.3 Following the Crack — Absenteeism, Non-Teaching Duty, Deployment

A sanctioned, filled teaching post is not the same as a teacher who is present and teaching. Kremer, Muralidharan, Chaudhury, Hammer and Rogers' foundational study used surprise visits to a nationally representative sample of government primary schools and found that "25% of teachers were absent from school, and only about half were teaching", with absence ranging from 15% in Maharashtra to 42% in Jharkhand. [11]

A decade-later panel by Muralidharan, Das, Holla and Mohpal, spanning 1,297 villages, found absence had only fallen to 23.6%, with the fiscal cost of unauthorised absence estimated at $1.5 billion a year — and better monitoring found to be over ten times more cost-effective than hiring more teachers. [12]

Case Study: Teachers Pulled Off Teaching Duty by the State Itself

Section 27 of the RTE Act permits deploying teachers only for census, disaster relief, or elections — yet Booth Level Officer (BLO) and electoral-roll revision work routinely pulls teachers from classrooms for extended periods. In 2025-26, the Delhi High Court directed the Election Commission to ensure such duties did not impose an "unbearable" burden; the ECI itself conceded some field orders had wrongly implied teachers must work during teaching hours and were withdrawn. [13] [14]

The Allahabad High Court separately held that teachers should be the "last resort" for election duties, warning that without free time "quality of teaching will become stagnant." [15] In December 2025, Maharashtra's own education minister wrote to the Chief Minister asking that teachers be exempted from BLO duty because students "miss regular classes and face difficulties in completing the curriculum." [16]

Complication — deployment and structure: even present teachers may not be able to teach effectively given how schools are staffed. UDISE+ recorded 1,04,125 single-teacher schools in 2024-25, enrolling nearly 34 lakh students — one adult responsible for every grade in the building. [17] [1] ASER 2024 found two-thirds of Class 1 and 2 classrooms in primary schools were multigrade. [2]

1.4 The Infrastructure Question

School infrastructure was, for two decades, the headline reform — and it worked, on its own terms. Per figures cited by the Education Minister from UDISE+, schools with electricity rose from 53% (2013-14) to 91.8% (2023-24); drinking water from 83.2% to 98.3%; and internet from 7.3% to 53.9%. [18]

The Mid-Day Meal — A Lever for Enrollment, Not Learning

Jayaraman, Simroth and de Véricourt, analysing 506,125 schools, found midday meals produced "a statistically significant 6.6% increase in primary school enrollment", with Grade 1 enrollment up 18% — but their learning finding is blunt: "midday meals are not associated with improved learning... the estimates are statistically insignificant and close to zero." [19]

Infrastructure improved dramatically over a decade while learning stayed roughly flat. The inputs India could most easily buy — toilets, wiring, meals — are precisely the ones least connected to whether a child can read. [3]

Complication (nuance): the relationship is not zero everywhere. Chakraborty and Jayaraman found midday meals can raise learning — but only where complementary teaching inputs (teacher attendance, usable blackboards) exist, not from infrastructure alone. [20] Infrastructure is a necessary floor, not a sufficient cause.

1.5 Who's Actually Teaching? — Recruitment Integrity in Government and Private Schools

In parts of India, the person hired to teach children was never qualified to do so. This is a two-decade documented record centred on Bihar, where roughly 2.5 lakh contractual teachers were appointed after 2005 based on submitted mark sheets rather than a competitive written exam — a design that invited fraud. [21]

The Vigilance Investigation Bureau scrutinised around eight lakh certificates from teachers recruited 2006-2015 — verifying degrees from 42 universities inside Bihar and 14 outside — before the 2026 dismissals, with salary recovery, with interest, also ordered. [24]

The Degree-Mill Supply Side: OPJS University, Rajasthan

In 2024, Rajasthan's Special Operations Group exposed OPJS University in Churu, alleged to have issued 43,409 forged and backdated degrees across 19 states and Nepal since 2013 — including unauthorised B.Ed and B.P.Ed teaching qualifications. [26] Investigators found the "university" had only 30 staff and no visible students; founder Joginder Singh Dalal and two officials were arrested, and the UGC barred fresh admissions. [27] [28]

The Private-School Side: Under-Qualification, Not Forgery

In February 2025, Himachal Pradesh's Elementary Education Department issued warnings to private schools to ensure TET and B.Ed/JBT compliance in hiring — an enforcement action that implies non-compliance had been occurring. [29] A related 2025 report found nearly 75% of TET aspirants in the state failed the exam outright, "exposing serious shortcomings in subject knowledge, conceptual clarity and pedagogical preparedness." [30]

Responsibility here is a chain, not a single villain: recruiters who selected on unverified paper, degree-mills that manufactured credentials, regulators slow to detect it, and private managements that hired outside qualification norms. [21] [26]

1.6 The Government's Own Policy Swings

Two policy choices sit at the centre of the accountability debate: automatic promotion, and curriculum pace. Section 16 of the RTE Act introduced "no-detention" — no child held back through Class 8 — to reduce dropout and exam fear.

In 2015, 23 of 28 states asked for the policy to be scrapped. [31] Parliament amended the RTE Act in 2019, and in December 2024 the Centre notified rules allowing detention in Classes 5 and 8 after a failed re-exam and two months of remedial instruction. [32] [33]

Complication: Fact: no-detention was rolled back. Inference (contested): that it caused the flat learning profile. The evidence for that causal claim is weak — learning was low and flat before, during, and after the policy. Detaining a child in a class taught above their level simply repeats the mismatch rather than fixing it.

The deeper mechanism is described in Pritchett and Beatty's "Slow down, you're going too fast": curricula move far faster than children learn, so students who miss foundational skills fall permanently behind. [34] A related study found many Class 7-8 children performing at only a Class 2 curricular level — a curriculum running five to six years ahead of the child. [35]

The government's constructive answer is Nipun Bharat (2021), a national mission targeting universal foundational literacy and numeracy by Grade 3 — the first systemic push to prioritise foundational learning over curriculum completion.

Part 1 establishes that teachers, infrastructure, recruitment, and policy each explain a piece of the puzzle — but not why schools with similar institutional metrics produce different outcomes. That anomaly forces the investigation outside the school building.

PART 2 — The Twist: What the System Cannot Fix Alone

2.1 The Question Nobody Puts in the Report

If teachers, infrastructure, and policy determined outcomes, states with comparable resources should produce comparable learning. They do not. State-level variation in Class 3 government-school reading in 2024 spans over 40 percentage points between the best and worst performers. [36] Uttar Pradesh, historically resource-constrained, improved government Class 3 reading from 6% (2014) to 27.9% (2024), overtaking wealthier states. [37]

A gap this wide, uncorrelated with resources, points to factors the education department does not control and rarely measures: what children bring to the school gate, what happens at home, whether they actually attend, and the social contract inside the classroom.

2.2 Before the School Gate — Early Childhood Readiness

FACT: children arrive at Class 1 with vastly different readiness. ASER 2024 shows pre-primary enrollment for 3-year-olds rose from 68.1% (2018) to 77.4% (2024), and for 4-year-olds from 76% to 83.3%, with Anganwadi centres the dominant provider. [2]

India is also correcting the age at which children enter Class 1: "underage" enrollment (age 5 or below) fell from 25.6% (2018) to 16.7% (2024). [38] In government schools specifically, this share dropped from 30.3% (2022) to 18.2% (2024). [39]

Complication (inference): pre-primary enrollment is not automatically pre-primary learning. An Anganwadi providing nutrition and custodial care but little cognitive stimulation raises the enrollment statistic without raising readiness — the same enrollment-vs-learning gap seen at the primary level, pushed one stage earlier.

2.3 The Home Half of Learning — Parental Literacy and Home Environment

FACT: a study of first-generation learners across three ASER waves post-RTE found "substantial learning gaps" between first-generation learners and other students that "have increased over the years." [40]

FACT: a Pratham/J-PAL randomised evaluation testing adult literacy classes for mothers and training in home-learning support found "significant but modest impacts on children's math scores" and increased maternal involvement. [41] A Bangalore cohort study of 940 mother-child dyads found parental education, maternal IQ, and a stimulating home environment (≥6 age-appropriate books, caregiver engagement) significantly associated with better outcomes. [42]

2.4 Enrolled ≠ Engaged — The Attendance Gap

FACT: near-universal enrollment coexists with far-from-universal attendance. ASER 2024 recorded rural government-primary attendance at 75.9% (up from 73% in 2022) against enrollment near 98%. [43] An academic analysis found roughly 30% of enrolled children absent on the day of the survey visit, ranging from 11% in Gujarat to 40% in Uttar Pradesh — describing this as "nominal" enrollment. [44]

What to avoid (UNRESOLVED motive): it is tempting to read the attendance gap as weak family "engagement." The measurable behaviour (absence) is clear; the motive is not. Absence can reflect illness, agricultural or domestic labour, distance, migration, or a rational judgement that a given school day offers little learning. The honest claim is behavioural, not attitudinal.

2.5 The Authority Question — Discipline, Corporal Punishment, and the Classroom Power Dynamic

The classroom's disciplinary régime changed fundamentally in 2009, and whether that change helped or harmed learning is genuinely contested. This section presents the debate; it does not adjudicate it.

FACT: Section 17 of the RTE Act, 2009 imposed an absolute prohibition on physical punishment and mental harassment in all recognised schools, building on a 2000 Delhi High Court ruling holding corporal punishment inconsistent with a child's dignity under Article 21. [45] [46]

The Child-Rights / Research Side

Corporal punishment persists despite the ban — one figure estimated over 225,000 students received corporal punishment annually in Delhi alone [47] — concentrated in government and rural institutions where teacher training and oversight are weakest. [48] International and Indian research generally associates it with fear, disengagement, and worse — not better — learning outcomes, the evidentiary basis for the ban.

The Traditional-Discipline Side (presented, not endorsed)

A qualitative Kolkata study documented that many teachers and some parents perceive the loss of physical discipline as having eroded classroom order, arguing the ban removed a tool without providing a workable substitute. [49] Research on teacher attitudes in Delhi and Bihar finds many retain "traditional views on discipline," continuing the practice partly from "societal acceptance of harsh methods" and the absence of an effective alternative. [50]

CONTESTED / UNRESOLVED: whether removing corporal punishment, without investment in alternative classroom-management training, left a discipline vacuum that itself impedes learning. Argued in good faith on both sides; the evidence does not cleanly resolve it.

2.6 When the Classroom Loses Control — Eroding Teacher Authority

There is a widely circulated narrative that students increasingly film teachers and threaten to post videos to get them into trouble. The verified evidence supports that such incidents occur and can end careers — but does not establish this as a systematic, widespread pattern. This section is explicitly flagged UNRESOLVED on scale.

→ A Bengaluru PU-college teacher was terminated after students filmed him beating a student and uploaded it. [51]

→ In Channagiri, Karnataka, a viral video showed students bullying and assaulting a Hindi teacher; he declined to file a police complaint fearing reprisal. [52]

Interpretation: most documented viral cases show cameras exposing teacher wrongdoing (assault, intoxication), which is closer to accountability than to authority-erosion. The "students filming to entrap innocent teachers" framing is real in anecdote but not demonstrated as a widespread phenomenon by any dataset located in this research. A documentary should present this as an open, under-measured question — not an established driver of the learning crisis.

PART 3 — What Actually Works

3.1 Teaching at the Right Level (TaRL)

The single most rigorously evidenced intervention for the Indian learning crisis is Teaching at the Right Level: assess children on basic skills, group them by learning level rather than age/grade, and teach foundational reading and arithmetic at that level for part of the day.

Developed by Pratham and tested with J-PAL researchers including Nobel laureates Abhijit Banerjee and Esther Duflo, TaRL has been evaluated in at least six RCTs in India (2001-2020), consistently improving learning when well implemented. [53] A widely cited estimate places the effect at roughly 0.28 standard deviations [54]; intensive "learning camps" in Uttar Pradesh produced gains of roughly 22-23 percentage points in the share of children able to read letters, words, and paragraphs. [55]

TaRL's evidence travelled internationally: Zambia scaled its "Catch Up" programme to 1,900 schools with Pratham/J-PAL support; Botswana, Nigeria, Madagascar, Niger and Côte d'Ivoire have adapted the model. [56]

Complication — why scale-up inside government is hard: TaRL is not a universal plug-in but an approach for reorganising how a system teaches. Government-delivery versions succeeded where monitoring and mentoring were strong (Haryana), and faltered where the design was too complex to implement uniformly. The binding constraints are bureaucratic: teacher buy-in, reliable monitoring, and protecting instructional time against the pull of curriculum completion.

3.2 The Post-Pandemic Surge

COVID-19 school closures were the largest natural experiment in the effect of not schooling — and the recovery from them contains the paper's most hopeful evidence.

Singh, Romero and Muralidharan's panel study of ~19,000 primary-aged children in rural Tamil Nadu found that children tested in December 2021, after 18 months of closures, showed deficits of about 0.73σ in math and 0.34σ in language versus identically-aged children in 2019 — but two-thirds of that deficit was recovered within six months of schools reopening, and a government-run after-school remediation programme contributed about 24% of the cohort-level recovery. [57]

Complication: not everywhere recovered. Jammu & Kashmir recorded ongoing learning losses of 2.3 points in reading and 2.1 in arithmetic in ASER 2024. [59] And absolute levels remain low nationally: even after the surge, only about one in four Class 3 children reads at grade level.

UP, Bihar, and Madhya Pradesh — not India's richest states — posted the largest gains, driven by focus and implementation, not spending. This is the empirical rebuttal to the 'resources determine outcomes' thesis.

3.3 What This Means for the System as a Whole

The three parts of this investigation converge on a single argument. Part 1 shows the visible, institutional failures are real but incomplete: teacher vacancies exist but the national PTR meets the RTE norm; teachers are absent and diverted to non-teaching duty; over a lakh schools have a single teacher; recruitment fraud placed unqualified adults in classrooms for two decades; and policy oscillated on detention while curriculum ran years ahead of the child. None of this explains why similarly-resourced states diverge by 40 points in reading.

Part 2 shows why: much of what determines learning enters before and outside the school — readiness gaps, home environment, the enrollment-attendance gap, and a contested, under-measured shift in classroom authority. These factors are largely outside the education department's direct control, which is precisely why input-based reforms improved the measurable inputs without moving the learning outcome.

Part 3 shows what does move outcomes: not more resources, but instruction targeted to the child's actual level. TaRL's randomised evidence, the rapid post-COVID recovery, and the surge in poor-but-focused states all point the same way.

India built a school system extraordinarily good at enrolling children and mediocre at teaching them, because it optimised for the things that are easy to count and easy to procure. The interventions that work succeed precisely by attacking the alignment problem rather than the input problem.

Recommendations for the Documentary's Evidentiary Spine

→ Open on the paradox, not a theory: near-universal enrollment against the ASER 2024 reading finding and the 2008-2024 decline. [1] [2] [3]

→ Structure Part 1 as elimination: too few teachers → refuted by national PTR; introduce absenteeism, deployment, infrastructure, and recruitment fraud as the real cracks. [1] [11] [21]

→ Use the Bihar timeline (2008 → 2015 PIL → 2026 dismissals) and OPJS University as concrete, named case studies for the accountability chain — avoid implying this is universal; it is state-specific. [21] [26]

→ Handle Part 2 with explicit on-screen epistemic labels (FACT / INFERENCE / UNRESOLVED). For parental literacy, foreground the resource-gap interpretation and reject the motivation-deficit framing. [40]

→ For corporal punishment and viral-video authority erosion, present both sides and state plainly that scale is not established by data. [49] [52]

→ End on TaRL, the Tamil Nadu recovery study, and the UP outlier — the actionable message: measure learning as seriously as enrollment, protect instructional time, and target instruction to level. [53] [57] [37]

Caveats

Government/primary data (UDISE+, ASER/Pratham) and peer-reviewed/NBER economics anchor every major quantitative claim. Secondary Indian journalism is used for case studies and should be treated as reporting of allegations/charges, not adjudicated guilt — Bihar FIRs and OPJS arrests are charges, not convictions. [24] [26]

The Himachal private-school under-qualification claim rests on an enforcement-warning story implying, not quantifying, non-compliance; a hard prevalence count was not located. [29]

The eroding-teacher-authority section is explicitly UNRESOLVED on scale — built from individual anecdotes, most of which show cameras exposing teacher misconduct rather than student entrapment. Do not present it as an established, widespread driver.

Causation vs. correlation is genuinely unsettled across Part 2 (parental engagement, discipline, attendance motives). This report labels these explicitly; the documentary should preserve those labels rather than flatten them into assertions.

Some figures (e.g., the 40-point interstate reading spread, the 25-year infrastructure-vs-learning contrast) come from independent analyst syntheses of official UDISE+/ASER data rather than the primary release itself. [3] [36]

Numbers shift by release cycle: single-teacher-school counts and teacher-vacancy totals vary by source and definition; where sources conflict, this report notes the range rather than forcing a single figure. [6] [7] [8]

Sources & References — 59 Sources

Every [n] tag in this document, and every Title below, is a clickable hyperlink. Ctrl+Click (Windows) or Cmd+Click (Mac) to open.

Research prepared for Vella Theory · September 2026 · 59 sources · Every [n] verified reachable · Research only — no narrative script. Story framing at author's discretion.