Retirement is a cash flow problem with three unknowns: what you will spend, how long you will spend it, and what the money earns above inflation. Everyone gets all three wrong in the same direction, because the spending is in the future, the years are uncertain and the return is an average of a path that has bad stretches. This module puts a number on each, with its source and date, and builds one worked plan with the assumptions on the page. It is education, not advice on your retirement.
5By the end you can
- Turn today's spending into a first year retirement need in future rupees, with an inflation rate you can date.
- Read the SRS life table for survival odds, and see why planning to the average lifespan is planning to run short half the time.
- Work the corpus formula by hand and see how sharply it reacts to inflation and to the number of years.
- Say what Bengen (1994) found, what the study did not claim about India, and what carries over from it.
- Read the Senior Citizens Savings Scheme rules and an annuity as a concept, and fit EPF, PPF and NPS around them.
What you will spend, in future rupees
Start from a number you know: what the household spends in a month on what continues after work stops. Rent or EMI may fall away; health spending usually rises. Say ₹60,000 a month, ₹7.2 lakh a year, in today's rupees. Spending in the first year of retirement is that figure grown by inflation to the retirement date.
| Inflation a year | First year need at 60 | The same, per month |
|---|---|---|
| 4% | ₹15.8 lakh | ₹1,31,467 |
| 5% | ₹19.1 lakh | ₹1,59,198 |
| 6% | ₹23.1 lakh | ₹1,92,428 |
₹7.2 lakh a year today, grown for 20 years. The need at 60 is more than double today's in every row.
Read one more paragraph, 83 words
The latest dated numbers: MoSPI's CPI release of 12 Oct 2026 (base 2024=100) puts September 2026 inflation at 5.54% year on year, provisional, with health at 1.40% and actual rent at 2.37% in the same release. The RBI's Monetary Policy Statement of 7 Oct 2026 projects CPI inflation at 5.2% for 2026-27. One month of health inflation says little about the next twenty years of medical bills, so the plan uses one rate for everything and shows what happens at 4% and 6%.
How long: what the life table says
The Registrar General of India publishes abridged life tables from the Sample Registration System. The report for 2019-23, dated September 2025, puts life expectancy at birth at 70.3 years and at age 60 at 18.4 years: 17.3 for men and 19.6 for women. A corrigendum to the report fixed a West Bengal figure and left the national tables as they were.
| Age | Persons | Men | Women |
|---|---|---|---|
| 65 | 90 | 89 | 92 |
| 70 | 77 | 74 | 81 |
| 75 | 61 | 57 | 66 |
| 80 | 43 | 39 | 48 |
| 85 | 26 | 23 | 31 |
The term this section turns on: Life table
Alive per 100 at age 60, from lx in the SRS abridged life tables 2019-23 (India, total): for persons, 79,434 at 60 and 21,030 at 85. Rounded to the whole number. The table ends at 85 and over.
Read 3 more paragraphs, 187 words
The average is the dangerous number. If you plan for 18 years, you plan for about half of 60 year olds to outlive the plan. The table gives the whole curve. Starting from 100 people alive at 60, the number still alive at each age is:
Half of the 60 year olds in this table are gone by about 78. One in four persons is alive at 85, and nearly one in three women. The table stops there. If I extend it past 85 with a simple fade whose average remaining life matches the table's 6.2 years at 85, about one in nine persons is alive at 90. That extension is my approximation, not a published figure. A plan to 90 is a plan for a 30 year retirement from 60, and the one I use below.
Two limits. The table is for a population, not for your family; good health, income and care shift individuals either way. And it describes people alive in 2019-23 and not the 60 year olds of 2046, who may live longer. Both point towards a longer horizon, not a shorter one.
The corpus, by hand
The corpus at retirement must fund a stream of withdrawals that rises with inflation. The present value of that stream is the withdrawal in year one times an annuity factor, where the factor uses the real return, the return above inflation.
factor = (1 - (1+rho)^-N)
/ rho x (1 + rho)
Corpus = W x factor
| Inflation | First year need | Real return | Corpus for 30 years |
|---|---|---|---|
| 4% | ₹15.8 lakh | 2.88% | ₹3.23 crore |
| 5% | ₹19.1 lakh | 1.90% | ₹4.42 crore |
| 6% | ₹23.1 lakh | 0.94% | ₹6.07 crore |
Illustration. Two points of inflation, from 4% to 6%, raise the corpus by 88%: the need grows and the real return shrinks at once.
First year need: the largest is 6% at ₹23.1 lakh
Read 3 more paragraphs, 144 words
Worked. W = ₹19,10,374, the ₹7.2 lakh grown at 5% for 20 years. After retirement the money earns an assumed 7% a year and inflation is 5%. So rho = 1.07 / 1.05 - 1 = 1.905%. For N = 30: (1.01905) to the power -30 = 0.5678, so 1 - 0.5678 = 0.4322; 0.4322 / 0.01905 = 22.69; times 1.01905 = 23.12. Corpus = ₹19,10,374 x 23.12 = ₹4.42 crore. The first withdrawal is 4.32% of it.
The figure is a multiple of 23 on a year's spending, and it is touchy. Hold the return at 7% and move inflation:
The same arithmetic with inflation at 5% and the return at 7% gives ₹3.84 crore for 25 years, ₹4.42 crore for 30 and ₹4.94 crore for 35. Five more years of life cost 12% more corpus. That is the longevity risk in rupees.
Safe withdrawal thinking, and what carries over to India
The rule of thumb people quote, "4%", comes from William Bengen's article "Determining Withdrawal Rates Using Historical Data", published in the Journal of Financial Planning in October 1994. What he did: for retirees starting in each year from 1926, he tested a first year withdrawal, followed by withdrawals raised with inflation, against actual United States returns, mostly with a portfolio of half common stocks and half intermediate term Treasury notes, and asked how long the portfolio lasted. His results: a 3% first year withdrawal lasted at least 50 years in every case he tested. A 4% withdrawal was never exhausted before 33 years in any past case, and 4.25% could have run out in as little as 28 years. For a client aged 60 to 65 he counselled about 4%, and called a rate of 6% or more gambling.
| Real return | 25 years | 30 years | 35 years |
|---|---|---|---|
| 0% | 4.00% | 3.33% | 2.86% |
| 1% | 4.50% | 3.84% | 3.37% |
| 2% | 5.02% | 4.38% | 3.92% |
| 3% | 5.58% | 4.95% | 4.52% |
| 4% | 6.15% | 5.56% | 5.15% |
First year withdrawal as a percentage of the corpus that lasts exactly N years when withdrawals rise with inflation and the real return is constant: 1 / factor from the formula above. At a real return of 0%, 30 years is 1 / 30 = 3.33%.
₹100 in Dynamic Asset Allocator became about ₹388 in the model
Read 2 more paragraphs, 176 words
I use it as context only, for three reasons. It is United States data: returns on American stocks and Treasuries, American inflation, a 1926 start. It describes the worst start year in that sample, which is one history. And it is a statement about a 30 year retirement under a fixed rule, not a law. The portion that carries over to India is the structure, not the number. A sustainable withdrawal depends on two things: the return above inflation and the years. Hold those and the arithmetic gives the rate directly:
Read across. At 2% above inflation, a 30 year retirement supports about 4.4%, near the 4% of the rule. At 1% above inflation it supports 3.8%. Where inflation is higher, the same nominal return leaves less real return, and a rule borrowed from American data can hand over a number that the local arithmetic does not support. Put your own inflation and return in the formula instead. The number that comes out is the withdrawal rate of the plan you wrote, with every assumption visible.
Sequence risk
The table above uses a constant return. Real returns arrive in an order, and in retirement the order matters, because money is leaving while prices fall. Take ten annual returns: -20%, -10%, 5%, 10%, 15%, 12%, 18%, 8%, 10% and 14%. A portfolio of ₹1 crore withdraws ₹5 lakh at the start of year one, raised 5% a year. Run the returns in that order, then in the reverse order. The ten numbers are identical in both runs.
| End of year | Bad years first | Bad years last |
|---|---|---|
| 1 | ₹76.0 lakh | ₹108.3 lakh |
| 3 | ₹61.1 lakh | ₹116.5 lakh |
| 5 | ₹62.9 lakh | ₹139.5 lakh |
| 10 | ₹62.9 lakh | ₹104.9 lakh |
Illustration with made up returns. Same ten returns, opposite order. The money runs out in year 19 in the first run and year 25 in the second, if returns are 7% after year 10 and withdrawals keep rising 5%.
Read one more paragraph, 139 words
Same average, a ₹42 lakh gap after ten years and six more years of life for the portfolio. Early losses hit a pot that is also being drawn down, and the withdrawals then lock them in. Bengen's own worst start years were 1937, 1946, 1969, 1973 and 1974, and he singled out the downturn of 1973 to 1974 as the most damaging because it came with high inflation: portfolios fell and what was left bought less. The defences are ones a household can use. Hold two or three years of spending in debt, so withdrawals do not come from falling equity; in the worked plan below two years of the first year need is ₹38.2 lakh. Spend less after a bad year. Take part of the income from a source that does not fall. Or work two years longer.
Annuities as a concept
An annuity exchanges a lump sum for a payment for life. A life insurer licensed by IRDAI prices it by pooling: those who die early leave money that pays those who live long. The price of ₹1 of annual payment is the present value of the payments, each weighted by the chance the person is alive to receive it.
| Payment form | First year payout per ₹100 |
|---|---|
| Level for life | ₹9.99 |
| Rising 3% a year | ₹7.83 |
| Rising 5% a year | ₹6.51 |
My calculation: SRS abridged life tables 2019-23, persons, linear between five year ages, exponential fade after 85, 7% discount rate. Not a market rate. Illustrates the cost of inflation protection.
Read 3 more paragraphs, 311 words
The SRS table lets me show the idea. Use the survival curve above, a discount rate of 7% a year, payments at the start of each year, and my extension of the table beyond 85. A level annuity for a 60 year old then costs about ₹10.0 for every ₹1 of yearly payment. That is a payout of about 10% of the lump sum a year, for men 10.4% and for women 9.6%, before any insurer cost or margin. It is a fair price under my assumptions, not a quote, and what an insurer offers on the day depends on its own tables, costs and yields.
The table is the point. The level payment looks high, and it buys less each year: at 5% inflation a fixed payment buys 61% of its first year value after 10 years, 38% after 20 and 23% after 30. Rising payments cost a lower start. An annuity swaps market risk for inflation risk and for the insurer's promise. Yaari (1965) showed that a consumer with an uncertain lifetime and no wish to leave money behind holds assets as annuities. Davidoff, Brown and Diamond (2005) relaxed the assumptions and found that positive annuitization remains optimal widely, though complete annuitization does not.
The National Pension System carries the idea into law. Under the PFRDA (Exits and Withdrawals under the National Pension System) Regulations, 2015, as amended on 16 Dec 2025 and read in the consolidated text, a non government subscriber who exits between 60 and 85 must put at least 20% of the corpus into an annuity (above the small corpus limits), and a government sector subscriber at least 40%. The annuity comes from a provider empanelled by the Authority. A further amendment in the Gazette of 14 Jul 2026 changes regulation 4A only, so these shares stand. Module 20.5 has the exit table by corpus size.
The Senior Citizens Savings Scheme
The scheme is a Government of India deposit for people from 60. The rules, as read in the Senior Citizens' Savings Scheme, 2019 and its amendments:
- Who. A person of 60, or one of 55 to 60 who has retired on superannuation or otherwise, if the account is opened within three months of receiving retirement benefits (the amendment of 9 Nov 2023). Defence personnel from 50.
- Limit. The maximum deposit was raised from ₹15 lakh to ₹30 lakh by G.S.R. 240(E) of 31 Mar 2023, with Department of Posts SB Order 06/2023. The limit applies across all accounts of one person.
- Rate. 8.2% a year for 1 Oct to 31 Dec 2026. The Department of Economic Affairs Office Memorandum of 30 Sep 2026 (F.No. 1/4/2019-NS) leaves the rates for that quarter unchanged from the quarter before, and the memoranda of 30 Jun 2026, 30 Mar 2026 and 31 Dec 2025 did the same. The Department of Posts leaflet of 7 Oct 2024 gives 8.2% from 1 Jan 2024. I did not read the memoranda for the quarters between, so the level of 8.2% rests on that chain and on that leaflet.
- Interest. Paid quarterly, on the first working day of April, July, October and January (rule 5(2)).
- Term. Five years, and the account may be extended in blocks of three years under the amendment published 9 Nov 2023 (G.S.R. 829(E)).
- Early closure. No interest if closed before one year (and any paid is recovered); 1.5% of the deposit is deducted if closed after one year and before two; 1% after two years.
Two terms to hold on to from 20.9
Read 2 more paragraphs, 154 words
The arithmetic: ₹30 lakh at 8.2% pays ₹2,46,000 a year, ₹61,500 a quarter. Against CPI inflation of 5.54% (September 2026) the real rate is 2.52%. The payment is fixed in rupees, so inflation eats it: at 5% a year, ₹2.46 lakh buys what ₹1.93 lakh buys today after five years, and the ₹30 lakh that comes back is worth ₹23.5 lakh. Interest is taxable as income. Under Schedule XV of the Income-tax Act, 2025 a deposit under the scheme is on the list for the ₹1.5 lakh deduction in section 123, which is old regime only.
For a retiree, the scheme is a steady quarterly income that covers a part of spending with no market risk. It is not a corpus. At ₹2.46 lakh a year it covers 13% of the first year need of ₹19.1 lakh in the worked plan. It is a bucket for the first few years, not the whole answer.
EPF, PPF and NPS in the picture
For most salaried people the retirement corpus is mainly the debt held in these three. Module 20.4 sets out how EPF and PPF work and how each rate is declared; Module 20.5 does the same for NPS. Here is how they sit in the arithmetic.
- EPF. 8.25% a year on accumulations for 2025-26, the rate the EPFO's Central Board of Trustees recommended on 2 Mar 2026 (Module 20.4 gives the approval and order). Against 5.54% CPI that is a real return of about 2.6%.
- PPF. 7.1% for 1 Oct to 31 Dec 2026, reset each quarter by the Ministry of Finance. Against 5.54% that is about 1.5% real.
- NPS. A market linked pension account whose exit forces part of the corpus into an annuity, as above.
Read one more paragraph, 85 words
Real returns of 1.5% to 2.6% are close to the rho of 1.9% in the worked plan, which is the point. Debt that earns about 2 points over inflation can sustain a withdrawal of near 4.4% for 30 years. The part of a retirement fund that needs to earn more than that, because the plan asks for it, has to be held in assets that can fall, and falls are where sequence risk lives. Module 4.6 works out how much equity a goal can carry.
A worked plan, with the assumptions written down
An illustration with made up inputs. A person is 40 and plans to retire at 60.
Where this section sits in module 20.9
The monthly figure is nearly as large as the spending it is meant to fund. That is not a calculation error. A household that spends ₹60,000 a month and wants to spend ₹1.59 lakh a month in 20 years needs to build a pot of 23 times the first year of spending. Three changes move the answer most:
- Retire at 65. The need at 65 is ₹24.4 lakh, but the horizon is 25 years: corpus ₹4.91 crore, existing money ₹2.05 crore, gap ₹2.85 crore over 25 years, ₹31,364 a month flat. Five more years of work cuts the monthly saving from ₹53,065 to ₹31,364.
- Inflation at 4% instead of 5%. The corpus is ₹3.23 crore, not ₹4.42 crore.
- A pension that does not fall. Whatever part of the need an annuity or the SCSS covers lowers the corpus the portfolio must carry, and removes that part from sequence risk.
Read one more paragraph, 53 words
The plan also hides things. The 8% before retirement and 7% after are my assumptions and are before tax. The plan ignores a house, a spouse's income, and the spending that falls with age or rises with illness. Each belongs in a review of its own, once a year, as Module 20.8 describes.
Key takeaways
- Spending at retirement is today's spending grown by inflation: ₹7.2 lakh a year today is ₹19.1 lakh in 20 years at 5%, and ₹23.1 lakh at 6%.
- SRS abridged life tables 2019-23 give 18.4 years of expected life at 60, but 26% of persons (31% of women) are alive at 85. Plan for a horizon longer than the average.
- Corpus = first year need x factor, where the factor uses the real return. At 7% and 5% for 30 years the factor is 23.12 and the corpus is ₹4.42 crore.
- Bengen (1994) found 4% survived every start year since 1926 in US data for at least 33 years. It is a US result. In India, use the arithmetic with your own inflation and return.
- The same ten returns in a different order left ₹62.9 lakh against ₹104.9 lakh after ten years. Order matters once money is leaving.
- An annuity trades market risk for inflation risk. A level payment starts higher and buys less every year.
- The Senior Citizens Savings Scheme pays 8.2% to a ₹30 lakh limit, ₹2.46 lakh a year, which covers 13% of the worked plan's first year need.
- Retiring at 65 instead of 60 cut the flat monthly saving in the worked plan from ₹53,065 to ₹31,364.
Questions readers ask
Is 4% a safe withdrawal rate in India?
The 4% comes from US data in Bengen (1994). The arithmetic here shows that a 30 year retirement supports about 4.4% if the money earns 2 points above inflation, and about 3.8% at 1 point. The right rate is the one that comes out of your own inflation, return and years.
How long should the plan last?
Longer than the average. Expected life at 60 is 18.4 years in the SRS table for 2019-23, but 26% of persons reach 85 and the tail goes further. A 30 year plan, to 90, is a common choice, and it is a choice, not a fact.
Is the Senior Citizens Savings Scheme enough on its own?
For most households it is a part of the answer. At 8.2% on the ₹30 lakh limit it pays ₹2.46 lakh a year, fixed in rupees, so its real value falls each year, and the interest is taxable.
Does an annuity solve longevity risk?
It removes the risk of running out, for the amount annuitised, and adds inflation risk if the payment is level. The pooling idea is old (Yaari, 1965). What a given insurer offers depends on its own pricing, which this module does not quote.
Rules and figures in this module were last verified on 31 Oct 2026. Tax and regulatory rules change; check the source before acting.
