The 4% rule in India: how big a corpus lasts for life

It comes from US research in the 1990s (William Bengen, then the 'Trinity study'): a mix of US shares and bonds, from which 4% a year rising with inflation survived every 30-year period in US history.

India is different. Inflation has been higher (we assume 6% a year), so the withdrawal grows faster, and debt returns are only a little above inflation. Both make a given rate run out sooner.

• Yearly spending means what your investments must pay: your expenses in retirement, less any rent or pension. Not your house, not your income.
• 20× (5% a year): lasted 25 years in 89 of 100 cases. A floor, not a target.
• 22× (4.5% a year): lasted 25 years in 95 of 100. Suits retiring at 65 or later.
• 25× (4% a year): lasted 30 years in 95 of 100, but 40 years in only 81. Suits retiring at 60 to 64.
• 28× (3.6% a year): lasted 35 years in 95 of 100. Suits retiring at 55 to 59; about 30× from 50 to 54.
• 33× (3% a year): lasted 50 years in 95 of 100. Enough for life at almost any retirement age: a sensible 'freedom number'.
• 40× (2.5% a year): after 50 years the corpus was still worth what it started with, in today's money, in 86 of 100. This is the number for passing wealth on.

4,000 sequences of 50 years built from actual monthly returns of Indian equity, debt and gold funds, 2007-08 to 2026-10 (2008 and 2020 included), in random 12-month blocks. Portfolio: 35% equity, 5% gold, 60% debt, rebalanced monthly. The withdrawal rises 6% a year. Fund returns are after fund costs, before tax.

A plan that says 'your money lasts until 82 in 4 out of 5 markets' asks you to bet on how long you will live and on what markets will do. A target asks a simpler question: have you built 25 times what your investments must pay in a year? If yes, the withdrawal is a small enough share of the corpus to survive bad decades; if not, you know exactly how far there is to go.

So retirement is planned to a number, set by how long the money must last (to 90): 22× for retiring at 65 or later, 25× at 60 to 64, 28× at 55 to 59, 30× at 50 to 54, 33× to be free at any age, 40× to pass it on. 20× is a floor, not a target.

• Start with the gap in rupees a month. That is the number to work towards, not a reason to give up.
• Grow the income: ask for the raise, change jobs, learn the skill that pays more, build a second income. Over a career this moves the target more than any fund choice.
• Raise your SIP with every increase in income, even by 5% a year.
• Cut the costs that don't matter to you, and keep the ones that do.
• Working two or three years longer both adds to the corpus and shortens the years it must pay for.

More equity makes the corpus grow faster but swing harder. At 60% equity, 4% a year lasted 50 years in 74 of 100 cases (against 68 at 35%), but early falls hurt more. Most retirees are better served by a modest equity share and a lower withdrawal rate than by a high equity share and a high one.

• Skip the inflation increase after a year in which the portfolio fell. It costs little and adds years.
• Keep two years of spending in liquid and arbitrage funds, so a fall never forces you to sell equity.
• Count rent, pension and an annuity first: they shrink what the corpus has to pay.
• Keep health cover going after you stop work. Medical costs are the most common reason retirement money runs out early.
• Review every year. A rule of thumb is a starting point, not autopilot.