Money after retirement

What the constraint actually is

After you stop earning, the portfolio stops being judged on its return and starts being judged on a payment. It has to produce a specific amount, on specific dates, for a length of time nobody knows. That is a different problem from the one every earlier article was solving.

Every plan written for the accumulation years carries a repair mechanism it never names: next month's income. Pick the wrong scheme, underestimate spending, meet a market fall three years in — each of those is absorbed by contributing again next month, and by the years still available for the contribution to work. Take away the contributions and the repair channel closes. Nothing about the arithmetic changed; the correction mechanism did.

So it is worth writing down the corrections that remain, because the honest list is short. Spend less, permanently or for a while. Sell more of the corpus, which brings the end date forward. Convert part of the capital into a contracted income, which an annuity does and prices accordingly. Return to some paid work, which is available to some people and not to others and rarely on demand. That is the whole set. The accumulation years had the same list plus three more — contribute more, contribute for longer, start the contributions again after a bad decision — and every one of those three was a claim on income that has now stopped.

The second half of the change is easier to miss. During accumulation the portfolio was measured against a long-run return, and a return is a single number with no dates in it. A drawdown portfolio is measured against a schedule of dates, and a schedule can fail in a particular year while the long-run number is still perfectly respectable. The rest of this article is about the ways that gap opens.

One thing that does not change: the money is not all being spent this year. Some of it is being spent in year 25, so the portfolio still carries a multi-decade job even though the person holding it has stopped earning. How large the corpus needed to be is the subject of how much you need to retire; what follows assumes the corpus exists and asks what running it down actually involves.

The same risk, with the sign reversed

The idea that the order of returns can decide an outcome the average cannot — sequence risk — is usually introduced as a retirement problem, which makes it sound like a new danger that appears at sixty. It is not new. It was present for the whole of the accumulation phase, and it was working in your favour.

Take ten illustrative annual returns — five years of −4%, then five years of +18% — and their reverse. They are the same ten numbers, so they compound to the same 1.8654 either way, which is 6.4% a year. Start with ₹1 crore and touch nothing, and both orders end at ₹1.87 crore. The order is not merely unimportant here; it is arithmetically incapable of mattering.

Now add a cashflow of ₹6 lakh at the end of each year, and run it twice: once as a withdrawal, once as a contribution.

The two gaps are the same size, and that is mechanical rather than a coincidence. The part of the final balance that comes from the opening ₹1 crore is identical in both orders, so the entire difference comes from the cashflow term — and that term is proportional to the cashflow, so reversing its sign reverses the answer without changing its magnitude. A bad decade at the start is a gift to somebody buying and a wound to somebody selling. The sign of the cashflow sets the sign of sequence risk.

The mechanism underneath is that a withdrawal in a fallen market sells more units to raise the same rupees, and those units are not available to participate in the recovery. The unit arithmetic of a withdrawal plan sets that out in full. What matters for the constraint in this article is the second-order effect: a 20% fall turns a ₹6 lakh withdrawal on ₹1 crore into a ₹6 lakh withdrawal on ₹80 lakh, so the rate being drawn rises from 6% to 7.5% at the exact moment the portfolio can least support it, and the balance now needs 25% to get back to level while still paying out.

Which is why the useful question is not "what return do I need" but "what does this plan do from a bad start". The largest first-year withdrawal that would have survived the worst historical start dates is the subject of withdrawal rates and sequence risk, and the structure most often used to blunt a bad start is the bucket approach. Neither removes the risk. Both change what a bad start is allowed to force you to sell.

Where a deposit portfolio keeps its volatility

The reason a deposit book feels different from a market portfolio is that its capital value never moves on a screen. That is a real property and it is worth something. It is also the reason the risk it does carry goes unnoticed for years at a time.

A deposit does not fix a rate. It fixes a rate for a tenor, and then hands you the proceeds on a date you did not choose the conditions for. Take an illustrative ₹50 lakh earning 7.5%: that is ₹3.75 lakh a year of income. If the money matures into a market where the comparable rate is 6%, the same ₹50 lakh now pays ₹3 lakh. Nothing was lost. No statement records a fall. The income simply dropped by 20% and stayed there.

So a deposit portfolio is not free of volatility. It has moved the volatility to the income line, where no chart exists and no app sends an alert. During the earning years this barely registered, because deposit interest was a side dish and a rate cut meant a smaller side dish. When the deposit book is the income, a repricing is a pay cut. The exposure has a name: reinvestment risk — the risk that you get your money back and cannot replace the terms.

If evidence is wanted that the rate is not yours to keep, the small savings schemes publish it themselves. Their rates are notified quarterly by the government: for the quarter to 30 September 2026, the Senior Citizens' Savings Scheme rate is 8.2% and the Post Office Monthly Income Scheme rate is 7.4%. Post Office time deposits show the term structure in the same table — 6.9% for one year, 7.0% for two, 7.1% for three and 7.5% for five. A deposit opened next quarter takes next quarter's number, whatever it is. That is a statement about how the rates are set, not a forecast of where they go.

Laddering — splitting the book so that a portion matures each year — is the standard response, and it is worth being honest about what it does. It does not remove reinvestment risk. It averages it, so that the whole income cannot reprice in one year, and it costs you the highest rate available at any single moment in exchange. That is a genuine trade and it is a reasonable one. It is not protection.

A drawdown-phase deposit book also runs into a ceiling that a smaller balance never touched. Deposit insurance covers ₹5 lakh per depositor per bank, principal and interest together, aggregated across branches and across every account held in the same capacity — and each capacity is insured separately, which is a narrower rule than it sounds. A deposit book small enough never to approach that ceiling raises no question at all; a book large enough to fund thirty years of payments, sitting in two or three banks, is a different arithmetic, and how the cover actually aggregates becomes a live question rather than a curiosity.

Why an all-deposit portfolio is not the safe choice over thirty years

Because a thirty-year retirement is not a three-year deposit held ten times over. The corpus is not one pot with one holding period. The rupee spent next month has a horizon of one month; the rupee spent in year 25 has a horizon of 25 years. A single allocation applied to the whole balance is answering thirty different questions with one answer, and it will be wrong at both ends: too exposed for the near money, too cautious for the far money.

That is the real argument for splitting a corpus by date rather than by product, which the bucket approach formalises and asset allocation generalises. It is also why "all deposits, because I am retired" and "all equity, because the horizon is long" are the same mistake with different signs. Each takes an answer that is right for one part of the schedule and applies it to all of it.

The table says one thing three ways. You are not choosing between risk and its absence; you are choosing which risk to hold and on which part of the schedule. The all-deposit book did not remove risk. It swapped a risk that announces itself for one that does not.

A second argument against an all-deposit book is the larger of the two over thirty years: a fixed income buys less every year. Inflation during retirement works that arithmetic properly and it is not repeated here. The point made above is narrower and holds independently of it — even with prices frozen, a maturing deposit still reprices, and the income still moves.

The first phase where you control the timing

Something genuinely improves after the income stops, and it is almost never listed among the compensations. For the first time, you decide when your taxable event happens.

Interest on a deposit does not wait to be spent. It is taxed as it accrues, every year, at your slab rate, whether it is withdrawn or left to roll over — the mechanics are here. An all-deposit book therefore generates a taxable event annually, sized by the book rather than by what you needed that year. A holding that grows instead of paying out is taxed when you sell it, and you choose how much to sell. The timing of the realisation moves from the bank's calendar to yours.

The arithmetic that follows is specific to a drawdown and rarely spelled out. Under the Income-tax Act 2025, long-term gains on equity shares and equity-oriented fund units fall under s.198 at 12.5%, and only above an annual ₹1.25 lakh; long-term means a holding period of 12 months, and anything shorter is taxed under s.196 at 20%. That exemption is annual and it renews. A retirement withdrawal is already a series of small annual sales rather than one large one, so it uses the threshold repeatedly where a single lumpsum sale would use it once. The shape of a drawdown happens to suit the shape of the relief.

Two corrections before that reads as a free lunch. Debt-oriented funds do not get the rate advantage: under s.76, units of a fund holding more than 65% in debt and money-market instruments, acquired on or after 1 April 2023, are taxed at slab rates, with the gain always treated as short-term. What survives there is only the timing — the tax still waits for the redemption. And a low total income does not shelter the gain: the new-regime rebate of up to ₹60,000 where total income does not exceed ₹12 lakh explicitly cannot be set against special-rate income such as an s.198 gain, which catches out exactly the reader whose other income is modest.

Now the reliefs that are specific to older taxpayers, with the regime named, because the regime decides whether they exist at all. There is a deduction for interest from deposits available to senior citizens, and it belongs to the old regime; a reader on the default new regime under s.202 does not get it. The same is true of the health-insurance deduction under s.126, which is ₹25,000, raised to ₹50,000 where the insured is a senior citizen — old regime only. Neither relief should be assumed into a comparison without first establishing which regime the return is being filed under, and the regime comparison is where that is worked through.

One structural note on that comparison. Several of the largest items on either side of it are attached to a salary and disappear with the salary, so a return built on interest and realised gains is running the same comparison on a different set of inputs and can land on the other side of it. Which regime suits a particular return is therefore not transferable from a colleague, a relative, or an article — including this one.

The plan has to be operable by someone other than its author

What follows is a property of the plan rather than of whoever runs it. A withdrawal plan is a running operation: maturities fall due, instruments are rolled, redemptions are placed, a bank account has to be funded before a standing instruction hits it. Most such plans can be run by exactly one person, and everything they need is in that person's head. That is a single point of failure.

It is a single point of failure at any age. What is different after the income stops is that there is nothing to bridge an interruption. A hospital stay, a period abroad, a lost password, a bank merger, a frozen account after a KYC lapse — while a salary is arriving, none of these stops the money. When the payment schedule is the income, a fortnight of nobody being able to operate the accounts is a fortnight with no money arriving, in a household that may have no other source.

Three things reduce it, and none of them is about markets. Fewer relationships: a plan spread across four banks, two brokers and a scatter of maturities is harder to run than the same money in half the places, and consolidation here is continuity rather than tidiness. A written operating note: what is held, where, what falls due when, and which account the money lands in — one page, kept current, and known to whoever would need it. And clarity about who is authorised to act, which is a question to put to each institution individually, because the answer differs and none of them volunteers it.

Keep two things separate that are routinely merged. Being able to operate an account while its holder is alive and being entitled to receive it afterwards are different arrangements. A nomination is not a transfer of ownership — what a nomination actually does is worth reading once, alongside how a will is prepared and estate planning. Accounts that exist only as logins need their own handling, which is digital assets planning.

Four errors that need an absent salary to be possible

None of these is a failure of discipline, and none of them takes this form while a salary is still arriving. Each is what a sound habit does once the constraint underneath it has changed.

1. Buying yield to close an income gap. The deposit book reprices, the income falls short, and the shortfall is met by moving to whatever is paying more. A higher promised rate is compensation for something — credit, tenor, liquidity or the strength of the promise itself — and after retirement there is no future income with which to replace the principal if the something arrives. The gap is real; the responses that do not put the capital behind it are on the spending side, or in which source the payment is drawn from.
2. Selling whatever is easiest. While a salary is arriving nothing has to be sold at all, so the question never comes up; in drawdown it comes up every quarter and it comes up under pressure, and the sale goes to the holding that is liquid, or up, or unembarrassing to sell. Repeated over a few bad years this quietly rebalances the portfolio into whatever performed worst, which is the opposite of what the plan intended. The fix is that the selling order is written down before it is needed, not chosen in the month it is needed.
3. The corpus as spare capital. A request arrives for a lump sum — a business, a property, someone you want to help, a joint venture with a relative. The corpus looks like the largest number in the household and therefore like the obvious source. It is not spare; it is already committed to a payment schedule, and the sum removed is not the cost. The cost is the income it was going to pay for the rest of the plan, which is the arithmetic the corpus article sets out.
4. Resting on more paid work. Deferring or resuming earning is the most powerful lever available before retirement and the least reliable one after it, because health, sector conditions and demand all remove it, usually at short notice. It is worth counting as an option that may not be exercisable rather than as a step in the plan — the same caution early retirement runs into from the other direction.

A fifth is not an error so much as an omission: the medical cost that arrives without a date. It cannot be scheduled and no allocation makes it safe, which is what cover is for. For indemnity individual health cover on someone aged sixty or above, premium revision is capped at 10% per annum, which matters because open-ended renewal pricing is the fear that most often stops a policy being held at all. What a claim involves in practice is in medical emergency planning, and the cover itself in the health insurance guide.

Testing the plan against the record rather than the average

Everything above runs on a constant illustrative return, which is a modelling convenience and not a property of any portfolio. The gap between the two matters more in drawdown than it ever did in accumulation, for the reason the sequence table shows: the average path and the actual paths give different answers once money is leaving, and there is no second attempt.

FNOTrader's Mutual Funds app runs on the full history of published daily per-unit prices — net asset values, or NAVs — from the Association of Mutual Funds in India, around 34 million rows of them. It simulates a schedule of cashflows on any scheme and period and reports the return on cashflows that land on irregular dates — XIRR — along with invested against value, the deepest fall from a previous peak along the way, and rolling-return distributions across every available start date. Two tests follow directly from this article: what a withdrawal schedule of the size you need would have done starting from the worst dates in the record rather than the average one, and how deep the fall was that a near-term sale would have had to be made into.

Historical outcomes describe what happened, not what will happen; past performance does not indicate future results. FNOTrader is not a SEBI-registered investment adviser and nothing here is advice. What suits a particular household depends on facts an article does not have, and the errors that are hardest to reverse are collected separately in retirement mistakes.

Common questions

Continue reading

More in Money by Life Stage · App: Mutual Funds · Definitions: glossary · Free tools: calculators · All: every article