85 Is Not A Conservative Assumption

Obesity drugs are turning into heart medicine. A gene-editing therapy is on the market. In August, a cancer treatment built from a patient's own tumour worked in a large trial for the first time.

We tried to price all of it into an Indian retirement plan. In rupees. Today.

It's worth close to nothing.

The number that did change things came from somewhere else. India's own death registration data.

The Number In Most Plans Describes Newborns

Indian financial plans usually run on life expectancy at birth. About 70 years, on the Sample Registration System estimate.

That's the wrong input.

Life expectancy at birth averages everyone born in a given year, including the people who died at 2, at 19, at 41. Your client didn't die at any of those ages. She's 58 and sitting across the desk from you.

She has already cleared the ages where almost all of India's mortality improvement happened.

What you need is the conditional figure. Given that someone has reached a certain age, how many more years can they expect? India publishes it.

Read the implied-age columns downward. An urban man who reaches 40 expects to see 75. One who reaches 70 expects to see 82.7.

The number climbs with every year already survived, against a life expectancy at birth of 70.3.

Same country, same data, same year. The row you pick up moves the planning horizon by 12 years.

The Gains Went To The Young

Life expectancy at birth rose about 20 years across that period. At 60, the gain was 4 to 5 years.

That difference decides whether any of the medical news matters to a plan.

For most of human history, the problem was that people didn't live long. Global life expectancy ran around 32 years in 1900 and reached 73.2 by 2023.

The 1900 figure is low mostly because so many infants and children died, with famine, epidemics, and war on top. An adult who reached 40 could already expect to see their sixties.

India ran the same sequence, compressed into a shorter period and starting from a worse place. Malaria control from 1953. Smallpox certified eradicated in 1977. Universal immunisation from 1985. Polio-free in 2014.

Infant mortality fell from 146 per 1,000 live births in 1947 to about 25 today. Maternal mortality fell by 86% since 1990, against a global 48%.

India's 20-year gain at birth is large partly because the base was so bad. The country was the epicentre of the 1918 influenza pandemic, with estimates between 10 and 20 million deaths, roughly 5% of the population.

The plague from 1896 killed around 10 million over two decades. The Bengal famine of 1943 killed somewhere between 800,000 and 3.8 million. Almost all of that fell on infants, children and young adults.

Recommended for you

Readers also explored

Blogs

FMCG Sector in India: Key Trends and Economic Drivers

9 min read

Blogs

India’s Unemployment Rate in 2025

7 min read

Eggleston and Fuchs found that at the start of the 20th century only about 20% of the gain in life expectancy came after age 65. By the start of the 21st century, close to 80% did.

A decomposition of India's own life tables from 1976 to 2020 found roughly 78% of the gain came from mortality decline before age 50. About 7% came from ages 70 and above.

India removed the things that kill people young. Those are different things from what kills people old.

For a client who has already reached 50, most of the last century's improvement is banked and can't be claimed twice. Whatever comes next has to be earned at older ages, and that's harder than anything medicine has managed so far.

What Medicine Is Doing Now, And What It Has Actually Shown

Three strands matter for how long people live.

Obesity drugs are becoming heart medicine. The GLP-1 class began as diabetes treatment. Obesity opens the door to diabetes, heart disease, kidney disease and liver disease at once, so a drug that works upstream may delay all of them together.

The SELECT trial showed semaglutide cutting heart attack, stroke and cardiovascular death by 20% in people who already had heart disease. It pointed toward fewer deaths overall, though the trial's design meant that finding was never formally confirmed.

Cardiovascular disease is India's largest killer, so this strand has the most room to move the number.

Cancer treatment is becoming individual. On 19 August, Merck and Moderna reported that intismeran autogene, a personalised mRNA therapy built from a biopsy of the patient's own tumour, met its endpoints alongside Keytruda in resected stage 2B to 4 melanoma.

It's the first treatment of its kind to succeed in a large trial. What it showed is that the cancer came back less often. Whether patients live longer is a different question the trial didn't measure, and one that takes years more to answer.

Genetic causes are becoming editable. CRISPR moved from laboratory concept to approved therapy with Casgevy, cleared for sickle cell disease in 2023. The newer work aims at common conditions rather than rare ones.

VERVE-102 lowers cholesterol permanently from a single infusion by editing PCSK9, the gene that regulates it. If that holds up, preventive medicine changes shape, from a statin taken daily for decades to one intervention. It's in a few dozen patients so far.

Then there's artificial intelligence, where the gap between what's claimed and what's shown is widest.

That last row comes from a PLOS Digital Health audit published on 19 August 2026. Its authors put it plainly: regulatory approval has outpaced clinical validation.

There's a pattern under all of it. AI works best at the start of the pipeline, where the question is whether a molecule looks promising. Its advantage disappears at the point where the drug has to work in a person.

None of these has been shown to extend life. Preventing a recurrence, or removing a risk factor, is a different thing from adding years.

The trial that would settle it for the obesity drugs is running now. It's the first in the class designed to measure deaths from any cause rather than deaths from one disease.

Demographers disagree about how much further the gains run. One tradition points out that every predicted ceiling on human life expectancy has been broken, usually within 5 years of being announced. Another finds improvement slowing across the world's longest-lived countries for 3 decades.

That argument won't settle in time to help anyone retiring this year. So plan with the data that exists.

Two Corrections Make The Average Wrong For A Household

India's registration data answers the planning question directly, at every age you're likely to be sitting across from. Two things have to be applied to it first.

These are averages, so roughly half of any group exceeds them. A plan built to the expected age fails for half the people who use it.

And a household has 2 lives. A couple retiring at 60 is funded until the second death.

The woman's column runs longer than the man's at every age in the table. Hers at 60 is 80.7, and half of that group sees more. So the household horizon sits beyond the longer of the 2 individual numbers.

Where exactly it should sit is a judgement about risk tolerance. What the data settles is that 85 is a central estimate for a household, and it's being used as a conservative one.

What Each Extra 5 Years Costs

7 assumptions, stated so you can argue with them.

Retirement begins at 60. The household draws ₹1,00,000 a month in today's money, or ₹12,00,000 a year. Withdrawals are fully indexed to inflation, so every figure here is real.

The portfolio returns 2% a year after inflation, and after tax, roughly what 8% nominal delivers against 6% inflation. Withdrawals come at the end of each year. The corpus is exhausted exactly at the terminal age, leaving nothing.

No other income. No lump-sum expense. No flexibility to cut spending in a bad year.

Those last 2 matter most. Real retirees do adjust spending, which makes these figures conservative. They also face medical costs that arrive in lumps, which pushes the other way.

Extending the horizon from 85 to 90 costs about 15% more corpus at the point of retirement. To 95, about 28% more.

The Two Errors Are Not The Same Size

The cost of not extending it is the same money seen from the other end.

A household that plans to 85 and finds itself funding a surviving partner to 90 runs out at 85 with 5 years still to fund. Replacing that income at that point takes ₹0.57 crore in age-85 money, which is what the ₹0.35 crore set aside at 60 would have grown into.

What separates the two errors is recoverability.

Over-accumulating shows up in advance and can be unwound by drawing down faster once the horizon gets clearer. A hole discovered at 85 can't be filled. No income can be earned and no product bought at that point.

The same relationship shows up from the other direction. Our simulation work on safe withdrawal rates for Indian portfolios finds the sustainable rate rising with retirement age, from 5.6% at 60 to 6.2% at 65, purely because the money has 5 fewer years to cover.

Extend the horizon rather than shorten it and the rate moves the same way in reverse. That work also finds no universal 4% rule for India, since the sustainable rate depends on the horizon and the allocation together. Those rates assume a diversified portfolio returning well above the 2% real used above.

The Years Being Added Are Not Free Ones

Funding a longer plan without knowing what's in those years is half an answer.

WHO estimates for India in 2021 put remaining life expectancy at 60 at 16.5 years for women against a healthy life expectancy of 11.9, and 14.8 against 11.2 for men. A gap of 4.6 and 3.6 years.

Roughly a quarter of all remaining years at 60, spent outside full health.

Those WHO levels sit below the SRS figures used above because they're modelled all-India estimates rather than urban registration data. So use the proportion from them and the level from SRS.

Applied that way, an urban woman reaching 60 can expect around 5.8 years outside full health and an urban man around 4.6. They arrive at the end, rather than spread evenly across the years.

The cost of those years doesn't show up in the index. Medical inflation runs at 12 to 14% a year against CPI health at roughly 1.6%, and out-of-pocket spending is still around 39.4% of total health expenditure in India.

Households absorb the difference directly. The first decade after 60 is spending. The last is care, and care isn't funded the same way.

What To Do With This

• Change the planning age first. Moving from a single-life estimate near 85 to a last-survivor horizon for the household is the biggest single change available in most Indian plans, and it needs no view on medicine at all.
• Plan for 2 lives. The chance of a surviving partner at 90 is a different problem from a median age at death, and it's the version most clients have never been shown.
• Look at working longer before looking at the portfolio. For a client in their fifties who is behind, working to 65 rather than 60 removes 5 withdrawal years. That cuts the corpus needed at 60 by roughly ₹0.57 crore before you count a single extra rupee of earnings. Nothing in the portfolio competes with that.
• Put pooled longevity protection on the table. An advisory firm arguing purely for a larger corpus points, conveniently, toward a larger managed portfolio. The honest version of this analysis doesn't.
• Stop treating 4% as safe. The sustainable rate depends on the horizon and the allocation together. Both extremes fail over a long retirement: pure debt to inflation and pure equity to a bad sequence early.

Ahead of the October prints, our economic calendar tracks every scheduled release with dates, previous readings and consensus.

Appendix: The Indian Sequence, In Full

The prose above compresses this. It's here for readers who want the underlying record.