Health, Learning and Living Conditions

The complete report has seven parts and one 204-page master PDF. The research cutoff for this part is 3 October 2026. Every figure retains its own observation year. Page references in the original reading guide below refer to this volume’s printed pages within the master PDF; use the contents here to move through the web edition.

What progress feels like.

A longer life. An easier school day. A bill a family can pay.

A country can look successful in a chart and still feel difficult to live in. I wanted this part to ask about the things that reach inside a home: whether a baby survives, whether a mother gets care, whether a child can read, and whether there is enough money for ordinary needs.

I write on MisterLove because I like taking a complicated idea apart until it becomes clear. My interests in research, business and technology keep bringing me back to one question: what does a number actually tell us about a person's life? A bigger hospital count and a healthier population are related, but they are different things.

The Matrix India posts started this series. They made me want to put the earlier figure, the later figure and the original source on the same page. In this part I return to its AIIMS, medical-college and MBBS-seat rows, and then widen the picture to health, learning, poverty and living conditions.

I am using 2014 as a point of comparison, not pretending that the country's history began there. Many changes started earlier. The Union government, states, public workers, private institutions and families all helped shape the results. A before-and-after table alone cannot divide the credit between them.

I have kept the dates visible, including the awkward ones. If the latest report published in 2026 measures 2023-24, I say 2023-24. If a new blood-testing method interrupts an older series, I explain the break. I would rather give you an honest gap than a neat number that means something else.

You do not need to know the jargon. Read the figure, then the paragraph beneath it. That is where I explain what it counts, what changed and how to picture it in everyday life.

Lovepreet Singh
MisterLove

Sources: 1, 2.

Where to begin.

Each topic stands on its own. The dates travel with the figures.

If you are short of time, start with pages 6, 15, 20 and 24. They ask four direct questions: are fewer babies dying, are children better nourished, can children read, and how many people remain below a poverty line? Each question needs its own measure.

Part 1 covered the economy. Part 2 covered physical infrastructure and services. This volume follows the people who use those systems. Parts 4-6 will cover production and industry, institutions and the environment, and state differences and international rankings.

The national figures here are a broad report card, not a complete census of every problem. Mental health, disability, every disease, every course and every local service need more evidence than fits in one volume. A missing row does not mean the issue disappeared.

The research cut-off is 3 October 2026. The newest observation differs by topic. Source numbers at the foot of each page take you to the reference section; a source title there opens the original publication.

All six evidence volumes and the final comparison are complete. The master PDF collects them together.

Dates first. Numbers second.

A report released today may describe life several years earlier.

Three dates can sit behind one figure: when people were surveyed, which period the question covered, and when the report was published. NFHS-6 was conducted in 2023-24 and released in May 2026. Some of its questions look back at births or care before the interview. Calling every result 'India in 2026' would hide that.

There was no national NFHS round taken exactly on the day the government changed in 2014. The available rounds include 2005-06, 2015-16, 2019-21 and 2023-24. Where I use 2005-06, the earlier figure is much older than 2014. Where I use 2015-16, it is already after 2014. Neither date has been relabelled.

A financial year runs from April to March. So 2013-14 ends in March 2014. School and college data use academic years. A three-year maternal-death estimate combines observations to make a rare event easier to measure reliably.

One lakh is 1,00,000. One crore is 1,00,00,000, or ten million. A rate per thousand compares equal-sized groups: 40 deaths per 1,000 births means four per hundred, not 40 per hundred. A percentage describes a share.

Moving from 40% to 60% is a rise of 20 percentage points. Relative to the starting 40%, it is a 50% increase. These are two different calculations. The Matrix chart's increase divided by the later total is a third calculation; it is not ordinary growth.

Counts, access and results belong on separate lines. A seat is a training place. A card is enrolment in a scheme. A hospital admission is an event. A lower death rate is an outcome. They help explain one another, but cannot replace one another.

Finally, a change after 2014 does not by itself prove a change caused by one government. These pages describe the measured change and the relevant programmes; they do not invent a precise share of political credit.

Sources: 4, 6, 7, 8, 9.

People are living longer.

Life expectancy is an estimate from death rates, not a promised age.

The earlier five-year period ends before 2014. The later estimate uses 2020-24 mortality and was released in 2026. The national gain is 3.1 years. Another commonly quoted baseline, 67.9, belongs to 2010-14, a period that includes 2014. Changing the period changes the starting figure.

Life expectancy asks how long a newborn would live on average if the period's death rates at each age continued. It does not say everyone born then will die at 70.6. Many people live much longer; some die much younger. Fewer deaths in childhood can raise the average even without the same increase at every adult age.

The female estimate exceeds the male estimate by 4.1 years. The urban estimate exceeds the rural estimate by 3.8 years. National improvement therefore sits beside differences between groups. An average of 70.6 does not describe the conditions in every district.

The same report puts Kerala at 75.6 years and Chhattisgarh at 64.7 years in 2020-24. That 10.9-year gap is larger than the national gain across the two periods above. State comparisons receive fuller treatment in Part 6.

Longer life does not automatically mean more years in good health. Living with diabetes, disability or another long illness still affects everyday life. This indicator measures length of life, not freedom from illness.

The rolling national series moved from 70.0 in 2016-20 to 69.8 in 2017-21 before rising again. Overlapping five-year estimates smooth events, including the pandemic; they should not be read as separate annual measurements.

Statement 1 in the SRS life-table report supplies the matched historical and recent series.

Sources: 3.

Fewer babies are dying.

Keep the age limit and the denominator on the same line.

For equal groups of 1,000 live births, the infant rate fell from 40 to 24. That is 16 fewer deaths per 1,000 and a 40% reduction relative to the earlier rate. It is a comparison of rates, not an audited count of exactly sixteen babies saved in each real village.

The infant and under-five measures overlap. A baby who dies in the first month is included in the neonatal, infant and under-five rates. Adding 18, 24 and 28 would count some deaths more than once. They describe different age boundaries around the same problem.

Neonatal deaths make up roughly three quarters of the infant rate: 18 out of 24. This points attention to pregnancy care, delivery, early feeding and care for sick or premature newborns. It does not prove that one programme caused the whole decline.

In 2024 the rural infant rate was 27, while the urban rate was 17. A ten-per-thousand gap remains despite the national improvement. The national row can improve while some places continue to face much higher risk.

SRS means Sample Registration System. It estimates births and deaths from a large sample, using continuing recording and an independent survey. These rates are different from a hospital's tally or the number of deaths that reached an administrative reporting system.

Lower is better for these measures. But 24 per thousand still means a real risk before the first birthday. A lower national rate is progress, not a statement that every preventable death has been avoided.

The latest SRS observation used here is 2024, published in 2026. Current-only rows have no invented 2014 counterpart.

Sources: 4, 5.

Childbirth has become safer.

Deaths and care coverage are separate measures of the change.

The maternal mortality ratio, or MMR, counts deaths related to pregnancy or its management, during pregnancy or within 42 days after it ends, for every one lakh live births. It excludes accidental or unrelated deaths. The ratio fell by 80 per lakh, about 48% relative to the 2011-13 level.

The periods combine three years because maternal deaths are relatively rare in a sample. For 2022-24, SRS gives a national uncertainty range of 78 to 95 around the central estimate of 87. A small movement from 88 in the previous overlapping period to 87 should not be treated as a precisely measured one-unit breakthrough.

Institutional delivery means giving birth in a health facility. The very old 2005-06 figure is shown honestly; it is not the situation immediately before 2014. Coverage was already 78.9% in 2015-16 and reached 88.6% in 2019-21 before the latest 90.6%.

Antenatal care means care during pregnancy. In NFHS-6, 76.2% received the first visit in the first three months, and 65.2% received at least four visits. Starting early and maintaining visits are different measures; four visits alone do not tell us whether all needed tests or treatment were provided.

A facility birth can help a woman reach trained staff and emergency care, but the building alone is not enough. Staff, blood, medicines, referral transport and treatment quality matter. Maternal mortality measures the outcome; the care figures describe parts of the route towards it.

Maternal-care percentages refer to the NFHS-defined birth and mother groups, not to all women in India.

Sources: 5, 6, 7, 8, 9.

Families are having fewer children.

Fertility describes current birth patterns; population changes more slowly.

Total fertility rate, or TFR, is the number of children a woman would have if the current birth rates at each age continued through her reproductive years. It is not the actual number of children in every existing family. A rate of 1.9 is an average across different patterns.

SRS and NFHS both show lower fertility over time, but they are different surveys with different reference periods. Their latest estimates, 1.9 and 2.0, should stay in separate rows. Substituting one source for the other can create an artificial movement.

Around 2.1 is often called replacement-level fertility: roughly enough births for one generation to replace itself in a population with low mortality, without migration. Falling below that level does not make the population shrink immediately. Large younger generations can still produce many births even when each woman has fewer children.

The contraceptive row concerns currently married women aged 15-49 using any method. The NFHS-6 release puts it at 69.1%, compared with 66.7% in NFHS-5. It is not a measure of contraceptive use by all adults, nor does it establish that everyone can choose the method they want.

Education, child survival, income, later marriage and access to reproductive healthcare can all influence family size. The numbers describe the shift; they do not isolate a single cause. Supporting informed choice and care requires more than treating a lower fertility rate as a target on its own.

SRS 2013 is linked as a supplementary source under reference 05. The NFHS and SRS series are kept separate.

Sources: 4, 5, 7, 8, 9.

Protection reaches more children.

The newest vaccination percentage has a narrower comparison group.

The first row is the older basic-vaccination indicator, using a vaccination record or information from the mother. The second is the card-based comparison highlighted in the NFHS-6 release. Reading 43.5, 76.4 and 87.1 as one unchanged series would mix the evidence used to classify children.

Basic vaccination in the older measure covers BCG, measles and the required doses of polio and DPT-containing vaccines. Receiving any one vaccine is a different, easier threshold than completing the full set. New vaccines such as rotavirus also have their own coverage measures.

The card-based improvement is 3.3 percentage points. It describes that comparison group, not a verified 87.1% coverage rate for every child regardless of documentation. Keeping the label attached avoids a stronger claim than the release supports.

Tuberculosis, or TB, remains another major infection. WHO's 2025 report estimates an incidence rate of 187 new and recurrent cases per lakh population in 2024. The March 2026 ministry summary reports a 21% reduction from 2015 to 2024. The baseline is 2015, already after the comparison year.

TB incidence estimates how many cases occur, including cases that may be missed. Notifications count cases reported to the system. Better detection can raise notifications even while the underlying disease rate falls. Vaccination coverage and TB incidence tell us about different infections; neither measures the whole burden of infectious disease.

The NFHS-6 publication uses the words based on vaccination cards. That qualifier is essential to the comparison.

Sources: 7, 8, 9, 18.

More public money for health.

The latest spending report describes 2022-23, not this year's budget.

Government health expenditure includes Union, state and relevant local spending, including capital investment. The nominal amount is almost three times the earlier level. Prices and population also rose, so tripling rupees does not mean tripling care for each person.

The GDP row compares the old denominator series retained in the report's historical table. For 2022-23 alone, the newly rebased GDP gives 1.48% instead of 1.43%. That change comes from the denominator, not from a second health-spending total. A consistent rebased historical comparison is not invented here.

Out-of-pocket expenditure is money households pay directly for care, medicines and related services, without reimbursement. Its share fell by 20.8 percentage points. Picture Rs 100 of all recorded health spending: direct household payments accounted for about Rs 64 earlier and Rs 43 later.

This does not mean every family's medical bill fell by one third. A national share can fall because other spending grows; some households can still face large bills. The latest 43.4% is also above the 39.4% recorded in 2021-22, so the recent path was not a continuous decline.

The NHA estimates were released in May 2026. They are expenditure accounts, assembled from several sources, not a fresh 2026 household survey. States contributed about 63.7% of government health spending in 2022-23; the national total is not Union spending alone.

GDP means the value of goods and services produced. The rebasing qualification prevents a false old-versus-new comparison.

Sources: 11, 12.

A wider financial safety net.

A household, a card and an admission are three different counts.

The household survey row includes different health insurance and financing arrangements. It asks whether at least one member is covered. A covered household does not necessarily mean every member has the same benefit or that every health expense is paid.

Ayushman Bharat-PMJAY began in 2018. It provides a hospital-treatment benefit of up to Rs 5 lakh per eligible family per year, under the scheme's rules. The extension covers senior citizens aged 70 and above. Coverage is not cash handed to every family, and the ceiling is not a promise to pay every kind of medical cost.

The latest backgrounder counts over 45.5 crore cards. These are cards issued to people, not 45.5 crore families. Hospitalisations count treatment events: one person admitted twice may contribute two events. Neither count is the number of unique people made healthy.

The backgrounder lists 38,466 empanelled public and private hospitals and Rs 1.92 lakh crore of covered hospitalisations by June 2026. Empanelment means participation in the network; it does not establish that every hospital offers every covered treatment nearby.

Insurance can reduce the shock of a hospital bill. Routine medicines, travel, lost wages and care outside the covered arrangements can still cost a family money. The spending accounts on the previous page help show why broader enrolment does not mean direct household spending has disappeared.

Public insurance and health services existed before PM-JAY. 'Not launched' in 2014 is the correct entry for this particular scheme, not zero health protection before 2014.

Sources: 7, 8, 9, 13.

More places to train doctors.

The Matrix chart is close to a newer official count, with dates missing.

The ministry's August 2026 reply supports these pairs. Medical colleges rose by about 119%, MBBS seats by 172%, and postgraduate seats by 177%. An MBBS seat is an annual undergraduate training place. A postgraduate seat is a place for further medical training; neither is a doctor already practising today.

The Matrix chart says 387 to 844 colleges and 51,348 to 1,39,000 MBBS seats. The newer reply gives 846 and 1,39,864. The broad scale is supported by this dated source, but the chart's later totals should be updated rather than presented as the latest exact counts.

An older 2025-26 academic-year snapshot in the August healthcare backgrounder lists 818 colleges and 1,28,976 MBBS seats. That older snapshot cannot be used to reject the newer 1,39,000-scale claim automatically. Reporting dates and academic-year vintages explain why the official publications can show different totals.

The reply includes government and private medical-education expansion. It does not say all of the colleges are new public institutions paid for by the Union government. Some new colleges upgrade existing district or referral hospitals; their existing services should not be counted as if nothing was there before.

A larger intake matters, but teaching quality, clinical exposure, fees, graduation and the place where doctors later work also matter. National training capacity can rise while a rural facility still lacks a specialist. The next pages distinguish project expansion from functioning care.

The August reply calls the baseline before 2014 and the later count at present. A more exact baseline date is not added here.

Sources: 2, 13, 14.

Twenty-three is a project total.

Approved campuses are not all complete hospital systems.

AIIMS means All India Institute of Medical Sciences. Add the original Delhi institute to these 22 new projects and the headline total becomes 23. But the official dashboard puts only six of the new campuses in its fully functional category. The other categories describe different stages of opening.

The Matrix chart's eight-to-twenty-three pair can describe a broader project count. Before 2014 there was Delhi, six new AIIMS that the ministry says had become partly functional, and the Raebareli project approved in 2009. Counting all eight does not mean eight equally complete hospitals were treating patients.

Outpatient care means a patient visits without being admitted overnight. Inpatient care means admission. A campus can start classes or open these services while still expanding beds, departments or buildings. That is real activity, but it is not identical to a completed campus with its full planned range.

The newer projects broaden the geographical spread of specialist institutions. The six earlier projects also continued developing after 2014. Their approval, construction and fuller operation cross political periods, so a one-line count cannot assign all work to the date of inauguration.

For a person seeking care, the useful details are which services work, how far away they are, whether staff and equipment are available, and how long treatment takes. The 23 headline explains scale. The stage-by-stage table explains how much of that scale has reached different forms of operation.

The dashboard covers new AIIMS under PMSSY and excludes the original Delhi institute. Its categories are reproduced separately.

Sources: 2, 15, 16.

Care needs people and supplies.

A nearby centre becomes useful through the services it can deliver.

Primary care covers prevention, basic treatment, follow-up and referral. Ayushman Arogya Mandirs developed from the health-and-wellness-centre programme, often strengthening existing facilities. The total does not mean 1.86 lakh completely new hospitals were built since 2014.

Indian Public Health Standards, or IPHS, cover facilities, staff, equipment, medicines and services. The August reply reports completed self-assessments, with 64% scoring above half the requirements. Completing an assessment does not mean meeting every standard.

The self-reported assessment helps find gaps. It does not show that 64% of centres have everything they need. A matched pre-2014 result for this same assessment is not supplied, so none is invented.

A doctor, nurse or technician must be present when the service is needed. Essential drugs must be in stock, tests must be usable, and serious cases must reach a facility able to treat them. Adding a building or a sanctioned post does not automatically add all of those pieces.

States recruit staff and deliver services, with support through the National Health Mission and other programmes. College expansion can increase the trained workforce, while hiring and deployment depend on local decisions.

The backgrounder also counts over 540 crore visits over time. Visits are repeated contacts, not 540 crore different patients. Activity and care quality need separate measures.

These are administrative and self-assessment snapshots, not an independently observed 2026 household access rate.

Sources: 13, 17.

Nutrition improved unevenly.

Height, weight and feeding habits answer different questions.

The old round is 2005-06, not 2014. It provides a genuine earlier observation, but the long interval includes years on both sides of 2014. The intermediate round matters too: it shows that different problems did not move together.

Stunting means a child is unusually short for their age against the WHO growth standard. It is associated with longer-term disadvantage in nutrition, health and care. Wasting means a child is unusually thin for their height. Underweight means unusually low weight for age and can reflect both problems.

For stunting and underweight, the survey threshold is more than two standard deviations below the reference median. In plain language, it is a defined low end of the expected growth range. Severe wasting uses the more extreme three-deviation threshold. These are measured classifications, not a judgement made by looking at a child.

Between the last two rounds, stunting fell by 6.2 percentage points and severe wasting by 2.5 points. Underweight fell by only 0.3 points. Describing all three as equally large improvements would hide the weaker movement in underweight.

The latest stunting result is about 29 children in every hundred, and underweight about 32. A child can fall in more than one category; adding the percentages does not give the number of children with any nutrition problem.

Feeding also changed: children aged 6-8 months receiving solid or semi-solid food along with breastmilk rose from 45.9% to 59.5%. The release's 95.6% breastfeeding figure for babies under six months describes any breastfeeding, not exclusive breastfeeding.

The severe-wasting row is not the rate of all wasting. All figures concern children in the survey-defined age group.

Sources: 7, 8, 9.

Some problems became worse.

An older result remains older, even when the new report omits it.

Anaemia means too little haemoglobin, the blood protein that carries oxygen, under the survey's age- and pregnancy-specific thresholds. Iron deficiency is one cause; infection, other nutrient shortages and inherited conditions can also contribute. The figure is not simply the percentage of people who did not eat enough iron.

The comparison showed anaemia rising among children and women, by 8.5 percentage points for children. It describes 2019-21, not October 2026, and cannot tell us whether prevalence has since risen or fallen.

The August 2026 ministry reply explains why NFHS-6 does not provide a new directly comparable anaemia headline: the ministry plans to use an independent ICMR survey with a venous blood-testing method. The older NFHS estimates used capillary blood. A change in the test method can affect the result even without the same change in people's health.

No replacement national result is inserted here without a verified release. Its absence is not evidence that anaemia improved. When the new result becomes available, the sample, thresholds and testing method need to be checked before making an old-to-new claim.

Nutrition also includes excess weight. The older comparable women's overweight-or-obesity indicator rose to 24.0%. It uses body mass index of at least 25 and the survey's pregnancy and recent-birth exclusions. It is a screening measure, not a diagnosis of every individual's health.

Under-nutrition, anaemia and excess weight can exist in the same country or family. More food or higher spending alone does not guarantee a balanced diet or better health.

The figures are deliberately dated 2019-21. They are not presented as new NFHS-6 results.

Sources: 7, 8, 10.

Food support changed its price.

The right to subsidised grain existed before 2014.

The National Food Security Act, or NFSA, became law in 2013, building on an older ration system. Calling the earlier provision zero would erase existing support. The later change made the regular grain entitlement free.

Free distribution began in January 2023, with continuation for five years from January 2024, through 2028. It uses the PM Garib Kalyan Anna Yojana name; the pandemic's additional grain entitlement was a different arrangement.

Priority households receive an entitlement per person. Antyodaya Anna Yojana, or AAY, covers the poorest eligible households and has an entitlement per household. A four-person priority household is entitled to 20 kg; an AAY household's entitlement is 35 kg. Mixing people and households creates the wrong total.

The 81.35 crore figure is a coverage ceiling based on the 2011 population, not the number of people currently below a World Bank poverty line. The 2024-25 annual report records about 80.67 crore people covered at its report date. Neither number establishes that every listed person collected the full ration every month.

Grain support can reduce hunger and free some money for other needs. Cereals alone do not provide every nutrient needed by children or adults. Nutrition results on the previous pages remain relevant even where ration coverage is broad.

Eligibility, the ability to collect grain, quantity and quality all affect delivery. A legal entitlement does not establish that it was received.

The ceiling, reported enrolment and actual grain collection must be kept separate. Figures here do not estimate the current poverty rate.

Sources: 19, 20.

Most children enter school.

Staying through the later years is a different challenge.

ASER visits households in rural India. Its enrolment question includes children in different school types, and its out-of-school group includes those who never enrolled and those who left. These results do not cover urban India. The 2014 survey was taken after the change of government, so it is a near-2014 starting point, not a strictly before-2014 observation.

Enrolment among younger children was already high. It rose by 1.4 percentage points across the decade, while out-of-school rates among 15-16-year-olds roughly halved. This helps explain why entering school and continuing later need separate lines.

UDISE+ 2025-26 covers 14.67 lakh schools, 24.72 crore students and about 1.03 crore teachers: institution-reported stocks, not ASER household estimates. A smaller student count would not automatically mean more dropouts; cohort size, reporting coverage and duplicate removal also matter.

In the NEP structure, middle school is Classes VI-VIII and secondary is IX-XII. The latest gross enrolment ratios are 89.6% and 71.7%. Gross enrolment divides all students in a stage by its expected-age population, including some younger or older students.

The new IX-XII stage cannot be compared directly with an old IX-X ratio. Its 51.9% retention rate measures progress through a school sequence; it does not mean 48.1% left in one year.

Using the same NEP stages, annual secondary dropout fell from 8.2% in 2024-25 to 7.0% in 2025-26. Middle-stage dropout moved from 3.5% to 3.6%. Progress therefore varied by stage.

The latest ASER basic national rural observations verified here are for 2024. UDISE+ observations use academic year 2025-26.

Sources: 21, 22, 23, 24, 25, 26, 27.

More schools have basic facilities.

A reported facility still needs to be usable every day.

This is a recent comparison using the same release's table. It is not relabelled as 2014 versus 2026. The biggest increases here concern computers and internet; both still fall well short of universal coverage.

For a group of a hundred schools, the reported computer-access share rose by about 22 schools and the internet share by about 18. These are percentage-point changes. They do not mean each school gained the same number of devices or every student can use them regularly.

The report also records drinking-water facilities in 99.5% of schools and handwashing facilities in 96.9%. A national entry for a facility's presence is not a continuous check of water quality, supply, maintenance or whether children can use it when needed.

Ramps with handrails reached 58.2%. That measure describes one accessibility feature, not complete access for every child with a disability. Entrances, classroom space, toilets, learning materials and support can still affect participation.

Teacher distribution matters alongside the buildings. The national report counts 1,00,843 single-teacher schools in 2025-26, down from 1,18,190 in 2022-23. It also records 5,663 schools with zero enrolment. A large total teacher workforce does not remove these local gaps.

National ratios can improve while a school still has a missing teacher or broken toilet. Reported resources do not establish teaching quality or successful learning.

The matched table begins in 2022-23. It avoids mixing older school categories or coverage rules with newer ones.

Sources: 21, 23.

Enrolment is not learning.

ASER uses simple tasks to check what children can actually do.

These are the report card's combined government-and-private-school trend figures. Class III means the third school year. The reading task uses a short text pitched at Class II level. Passing it establishes that task, not complete mastery of every part of the child's course.

The intermediate years matter. Reading results were lower in 2022, after the disruption of the pandemic, and recovered in 2024. Class V reading reached 48.8%, slightly above 2014 but below 2018's 50.5%. Saying learning simply rose steadily over the decade would hide that path.

For Class V, 48.8% means just under half could read the lower-grade text in this combined group. More than half could not complete that task at the required level. This does not mean they could read nothing, nor is it a label for their intelligence. It identifies a learning need.

Arithmetic improved more over the two endpoints. Class III subtraction rose by 8.3 percentage points and Class V division by 4.6 points. The reading and arithmetic results should stay separate: improvement in one does not prove the same improvement in the other.

ASER assesses children at home, so it can reach enrolled children absent from school that day. It covers rural India and uses basic tasks, not a national board examination. Its results should not be presented as an urban-and-rural national pass rate.

Income, language, prior learning and school choice also affect results. A public-private gap alone cannot prove that school type caused it.

Some all-children tables differ slightly from the combined government/private trend because their covered school groups differ.

Sources: 24, 25, 26, 27.

More access to college.

The latest national college survey measures 2023-24.

AISHE is the All India Survey on Higher Education. The July 2026 release measures 2023-24. Its comparison starts in 2014-15, already after 2014. Enrolment grew by about 1.08 crore, or 31.5%. This is a student stock, not graduates added in one year.

Gross enrolment ratio, or GER, divides all enrolled students by the estimated population aged 18-23. Some students are older or younger. Therefore 30% GER does not mean exactly 30% of 18-23-year-olds attend college.

Women's enrolment grew faster than the total over this period. A female GER of 31.2% exceeds the male GER, giving a gender-parity index above one. This describes participation relative to the respective age populations. It does not mean women already have equal representation in every course, senior academic post or later job.

The latest GER is 27.8% for Scheduled Castes and 22.8% for Scheduled Tribes, compared with 30% overall. Their earlier ratios were 18.9% and 13.5%. Access expanded, but the group differences remain; a larger overall student count cannot erase them.

Institutions voluntarily upload data to AISHE. The report describes validation and estimation, and warns that appearance in its list does not itself confer recognition. Enrolment also does not establish teaching quality, course completion or employment.

Expansion needs teachers, usable facilities, affordable courses and support to finish. The survey counts participation; learning and later use of qualifications need separate evidence.

AISHE GER uses the official projected 18-23 population; the report's method and response coverage remain relevant.

Sources: 28, 29.

More people can read and write.

The recent figure is a survey estimate, not a new census count.

Literate means able to read and write with understanding in at least one language. It does not require a college degree, fluent English or completion of a particular school year. The table uses people aged seven and above; including younger children would change the denominator.

The earlier result is the final 2011 census figure, not a 2014 measurement. The often-repeated 74.04% belongs to the provisional 2011 result. Using the final 73.0% consistently avoids quietly switching the historical series.

The recent result comes from the 2025 Periodic Labour Force Survey, whose annual report was published in 2026. It estimates literacy from a household sample. Comparing a census with a sample survey gives useful context, but is not a perfectly unchanged measurement system.

The difference between these endpoints is 8.1 percentage points. In the recent survey, men's literacy exceeds women's by 12.6 points, and urban literacy exceeds rural literacy by 11.6 points. Progress and remaining gaps can be true at the same time.

PLFS was redesigned for 2025, including changes to sampling and visits. A small change from an older PLFS annual figure should not be treated as a precise year-to-year gain without checking the design. The latest 81.1% is dated openly rather than replacing it with the older 80.9% commonly quoted for 2023-24.

Reading a medicine label, understanding a form or spotting an online scam needs further skills. Basic literacy alone does not measure these. Adults learning later in life also deserve clear information.

PLFS 2025, Appendix Table 7, rural plus urban: age seven and above. Cross-source differences require care.

Sources: 30, 31.

Budgets and family costs differ.

More budgeted rupees do not automatically buy proportionately more learning.

The first row compares the ministry allocations announced in the two Union budgets. The nominal amount is about 2.11 times the earlier amount. It is a budget estimate for the year, not final money spent, and not the whole country's education expenditure.

States finance much of schooling. Other ministries also spend on education-related activities. A claim about total public spending as a share of GDP needs a combined, consistently defined series. Dividing the Education Ministry's allocation alone by GDP would answer a narrower question.

Prices, student numbers, salaries and the range of funded work changed between the two years. The 111% nominal rise does not establish a 111% rise in real resources per child. Actual expenditure and what it delivers must be checked separately.

The second row comes from households interviewed in April-June 2025. It estimates school-education spending per enrolled student during the current academic year under the survey's rules. It is not the government's cost per student or a fee that every family is charged.

The non-government category combines different school arrangements; it is broader than private unaided schools alone. Different student backgrounds and school levels contribute to the average. This cost gap therefore cannot be treated as a clean price comparison for two otherwise identical schools.

The same survey puts 55.9% of enrolled school students in government schools, including 66.0% in rural areas and 30.1% in urban areas. Public provision remains central to access. A family's decision also depends on distance, teaching, language, fees and the cost of staying in school.

The budget row is in current rupees and compares announced ministry allocations. The household row is a separate 2025 survey.

Sources: 32, 33, 40.

Poverty fell. The line matters.

Use the same poverty line and price basis at both ends.

The April 2026 India Development Update gives these estimates using the 2011-12 consumption survey and HCES 2023-24. The observation dates are not 2014 and 2026. Under the lower line, the share fell by 24.5 percentage points; under the higher line, by 41.6 points.

PPP means purchasing power parity: an adjustment for what money buys across countries. The dollar figure is a common purchasing-power unit. Multiplying $3 by today's rupee-dollar exchange rate would produce the wrong Indian threshold.

The lower threshold counts extreme deprivation. Crossing it does not mean a household can comfortably pay rent, healthcare, education or an unexpected bill. At the $4.20 threshold, the report still estimates about 23.2 crore people below the line, compared with about 3.5 crore below $3.

The World Bank revised its international lines in 2025 using 2021 PPP. Old headlines using $2.15 and 2017 PPP cannot be put beside these figures as if the threshold never changed. A changed poverty line can alter the percentage without the same change in people's lives.

Indian consumption surveys also changed their questionnaires, visits and the periods people were asked to remember. The Bank constructs a welfare measure from food and non-food spending. Comparability involves statistical choices; it is not just counting bank balances.

The decline is meaningful under the stated method. It does not identify one cause or prove every family is secure. Poverty measures and food-support eligibility count different things.

This is the World Bank monetary-consumption measure. It is separate from India's older national poverty line and from MPI.

Sources: 34, 37.

A second way to measure hardship.

Some widely quoted before-and-after figures are projections.

Multidimensional poverty asks about several disadvantages together. NITI Aayog's national index uses twelve indicators across health, education and living standards. They include nutrition, child mortality, maternal care, schooling, fuel, sanitation, water, electricity, housing, assets and banking.

Each indicator has a rule for deprivation and a weight. People in a household are counted as multidimensionally poor when its weighted deprivation score reaches at least one third. This is not the share with any single problem, and it is not an income threshold.

The direct comparison comes from two surveys. It records a fall of 9.89 percentage points and an estimated 13.5 crore people leaving multidimensional poverty between 2015-16 and 2019-21. These are sampled household conditions, not the endpoints of the government's term.

The popular 2013-14-to-2022-23 pair uses extrapolation because surveys were not taken at those endpoints. The paper applies earlier rates of reduction to estimate the missing years. Its 24.82 crore figure is therefore an estimate under that method, not a direct survey count of people interviewed in those two years.

NFHS-6 was released in 2026. A new, comparable national MPI from it was not verified for this volume. The 2024 projection remains separate from the newer survey.

Money poverty and MPI can move at different speeds because they measure different things. A household can gain electricity or schooling and still have little money left after paying essentials. Reading both measures helps show why one poverty percentage cannot describe every kind of hardship.

The first row is survey-based. The second and third rows are estimated endpoints, not direct NFHS measurements.

Sources: 9, 10, 35, 36.

Households spend more on average.

Separate the rise in prices from the rise in real consumption.

Monthly per capita consumption expenditure, or MPCE, spreads a household's recorded consumption across its members. It includes more than wages spent that month. It is not salary, take-home income, savings or money left after bills.

At current prices, the rural average is about 2.88 times the earlier figure. At constant prices, the rise is about 45%. For urban areas, the corresponding real increase is about 38%. Removing the price rise makes the change in purchasing power easier to see.

The figures exclude the estimated value assigned to selected free benefits. Including those items gives a higher recent total: Rs 4,247 rural and Rs 7,078 urban. Both treatments can be useful, but switching between them in one comparison would change what is counted.

The survey's consumption Gini fell from 0.283 to 0.237 in rural areas and from 0.363 to 0.284 in urban areas between 2011-12 and 2023-24. A Gini nearer zero means consumption is more evenly spread under this measure. It is not a measure of land, total wealth or all income.

The average can rise while some households struggle. It is not the middle household's result, and very wealthy households are difficult to capture fully in household surveys. Lower consumption inequality does not establish that every form of inequality fell.

The survey design also changed between the rounds. These are the published comparisons, with that limitation kept visible. They offer evidence about material living standards; they cannot settle healthcare quality, debt pressure, unpaid work or access to good education on their own.

Constant-price figures use the report's 2011-12-price treatment. The survey period is 2023-24, not 2026.

Sources: 37, 38.

More homes can access LPG.

A connection and regular clean cooking are different achievements.

LPG means liquefied petroleum gas, supplied mainly through cylinders. Consumer accounts describe the supply network. They do not prove that every account is actively used or every family cooks exclusively with LPG.

Pradhan Mantri Ujjwala Yojana, or PMUY, began in 2016 to provide deposit-free connections to eligible women from poor households. Zero PMUY connections in 2014 means this scheme did not yet exist. It does not mean there were no LPG connections or earlier fuel programmes.

NFHS asks about household cooking fuel. Its clean category includes electricity, LPG or natural gas, and biogas. The 58.6% figure describes 2019-21, not 2026, and is a different measure from the current account count. A newer matching household-use percentage is not invented from the connections.

The refill average rose to 4.71 in 2025-26 for PMUY consumers. This is annual use under the programme's reporting, not 4.71 cylinders for every family. Family size, meals cooked, sharing and use of other fuels differ. The average alone does not establish exclusive LPG cooking.

The backgrounder also reports LPG coverage above 100% relative to its estimated household denominator. This administrative ratio can exceed a hundred because connections and household estimates are different counts. It cannot mean more than 100% of real households use clean fuel.

Water, sanitation, electricity and safe housing, covered in Part 2, also affect daily living. Their usefulness depends on regular supply, affordability and maintenance after the first connection or completion.

The newer refill result is FY2025-26. The NFHS clean-fuel observation is deliberately retained with its older date.

Sources: 7, 8, 39.

More women have useful access.

Having a tool and being able to control it are related, but not identical.

These indicators concern the women interviewed in the survey's age range, not every woman of every age. The earlier account and phone observations are from 2015-16; the internet baseline is 2019-21. None is silently moved back to 2014.

An account that a woman says she herself uses is a stronger measure than a bank's total number of opened accounts. It still does not tell us the balance, frequency of transactions or whether she can make every financial decision independently.

The bank-account measure rose by 36 percentage points, while the own-phone measure rose by 17.7 points. A household may have a smartphone that a woman cannot use whenever she needs it. Household access and personal access can therefore show different results.

Ever using the internet is a broad threshold. Trying it once qualifies; it does not establish daily use, a reliable connection, the skill to complete a form or control over a device. Its rise from 33.3% to 64.3% is substantial, but should not be described as universal digital independence.

Accounts and phones can help with payments, information, work and contact with services. The survey does not by itself establish that each woman gained all of these benefits or that every barrier disappeared. Affordability, confidence, language, safety and decision-making also shape what access becomes.

The figures throughout this part are national averages. They describe meaningful change while leaving differences between states, income groups, rural and urban areas, ages and communities visible. Part 6 examines those differences more closely. Part 4 covers farming, industry, science, space and defence.

Sources were checked through 3 October 2026. Each observation keeps its own survey year; later volumes remain planned.

Sources: 7, 8, 9.

Sources and reading notes