Money in your fifties

Why irreversible is arithmetic, not attitude

The constraint in your fifties is not that risk suddenly became more dangerous. It is that the instrument you have always used to reverse a mistake — the earning years still in front of you — is nearly spent. That changes which decisions can be taken twice.

Put a loss in the only units that make the point, which are years of saving rather than rupees. Take a household adding ₹9 lakh a year to its investments. A ₹25 lakh loss — a concentrated position that halved, a business that did not work, a property that would not sell — is 2.8 years of complete saving, because 25 divided by 9 is 2.8. At thirty-eight, with twenty-two years of contributions still to come, that is 13% of what remains. At fifty-six, with four, it is 69%. Same loss, same person, a different order of damage.

The mechanism underneath is worth stating flatly, because the popular version of it is wrong. Most financial mistakes are not repaired by a market recovering. They are repaired by the person earning and putting the difference back, with the market recovery doing part of the work on whatever survived. Recovery is usually the payslip, not the market — or the invoice, for anyone who bills rather than draws a salary — which is why the number of earning years left is the variable that decides whether a loss is a setback or a permanent reduction.

That does not make caution the answer, and the rest of this article is largely about why not. What it does is sort decisions by whether they have a door and when the door shuts. The ledger below is the version specific to this decade; none of these rows is live at thirty.

Read the last column rather than the first. What separates this decade from the previous two is not the size of the decisions but the fact that several of them are governed by closing doors rather than prices — and a closing door cannot be answered by a better return.

The opposite mistake, and what it costs

The conclusion people draw from all of the above is that a portfolio in its owner's fifties belongs in deposits. That reading has a price, and the price is calculable rather than arguable.

Keep the illustrative case fixed for the rest of this article: spending of ₹9 lakh a year in today's money, work stopping at sixty, and a retirement funded to ninety-five — thirty-five years, chosen as a planning assumption and not measured from anything. At an illustrative real return of 3%, meaning 3% after inflation, the corpus needed is ₹9 lakh multiplied by 21.49 — the multiple of one year's spending that funds thirty-five indexed years while the untouched balance keeps earning, which is the thirty-five-year annuity factor — and that is ₹1.93 crore. At a real return of zero — a holding that exactly keeps pace with prices after tax, which is the optimistic case for a taxed deposit — you need thirty-five times ₹9 lakh, or ₹3.15 crore. Refusing the growth asset costs 63% more capital, and it has to be found by someone with four years of contributions left.

The construction of that corpus figure, and the unit discipline that makes it ₹1.93 crore rather than a much smaller number, belongs to the corpus arithmetic and is not repeated here. What is different at fifty-six is that the two horizons in the calculation have pulled as far apart as they ever get: four to nine years of earning against thirty-five years of spending. This is two horizons, not one, and the shorter one is the only one that shortened.

Which is the same principle the forties article states as age not being a horizon, arriving at its extreme. The money spent in the reader's late eighties has a holding period of thirty years whoever is holding it today. For that money, deposit-like holdings are not the cautious choice — they are the one where erosion is the certain part. What that erosion does to a fixed income over decades is the whole subject of inflation after you stop earning.

Two mechanical notes that only bite at this size. Deposit interest is taxed as it accrues each year at slab rates, so the tax comes out of the compounding base annually rather than at the end — the comparison against a fund is set out in debt funds against fixed deposits, with the caveat that gains on units of a Specified Mutual Fund under s.76 of the Income-tax Act 2025 are taxed at slab rates, with the gain always treated as short-term. And deposit insurance covers ₹5 lakh per depositor per bank, a ceiling that only starts to bind once one bank holds more than that — rare for most of a saving life, routine for a consolidated retirement corpus — so how the cover aggregates now decides how many banks the money sits in.

None of which argues for holding the same allocation at fifty-six as at thirty-six. It argues for matching each holding to the date its money is spent, which is asset allocation, and which produces a more conservative position for the next five years and a more aggressive one for the money that will not be touched for twenty-five.

The home loan: rate, liquidity, and an option that expires

Where there is a home loan running into the retirement years, whether to clear it before stopping work is usually the largest single decision of this decade. It gets argued as a rate comparison. The rate comparison is the easy half.

Start with it anyway. Prepaying earns you the loan rate, with certainty, for as long as the loan would have run — there is no other place most households can put money and know the return in advance. What reduces that rate is any tax relief actually received on the interest, and this is where the regime matters more than the rate does: the new regime is the default under s.202 of the Income-tax Act 2025 and carries no relief for self-occupied home-loan interest, so for a taxpayer who has not opted out, the cost of the loan is the full rate. Under the old regime the relief exists and lowers it. Which of the two applies is worth settling before the comparison, not after, and the regime comparison is where that is done.

Then check what the exit costs, because it is often nothing. The current rule is no pre-payment charge, and no minimum lock-in, on a floating-rate loan taken by an individual for a purpose other than business — and the word doing the work is floating, which most home loans are and most unsecured personal loans are not. The distinction, and what to check on a specific sanction letter, is in foreclosure charges.

Now the half that is specific to this decade, and it is not about rates at all. Money put into a house stops being money. Getting it out again after employment income ends is a different transaction from the one available today: borrowing is assessed on income you no longer have, and the route that remains puts the home up as security — see loan against property. The option to borrow does not expire when the loan ends. It expires with the last payslip, and prepaying spends it early.

There is also an asymmetry in the loan's favour that almost nobody states, and it runs the opposite way to intuition. An equated monthly instalment is fixed in rupees. Everything else in a retirement budget is not — food, power, medical care and rent all reprice, which is why the corpus arithmetic indexes them. So the EMI is the one shrinking line in the whole budget. Put numbers on it: ₹40,000 a month is ₹4.8 lakh a year, and six years of it past the stop date, discounted at an illustrative 7% nominal because the payment itself is nominal, needs ₹22.9 lakh set aside. The same ₹4.8 lakh a year of ordinary indexed spending, discounted at the illustrative real 3% used above, needs ₹26.0 lakh. The fixed payment costs roughly 12% less to fund, for the same rupees, purely because inflation is working against the lender rather than against you.

Those two figures use different discount rates on purpose, and mixing them would destroy the point: a fixed nominal payment must be discounted at a nominal rate and indexed spending at a real one. Applying the real rate to the EMI would quietly assume the EMI rises with prices, which is the single thing it does not do.

The decision is also not binary, which is how it is almost always framed. Between clearing the loan and keeping it sits holding a reserve large enough to clear it and deciding later — which keeps the option open at the cost of the spread between the loan rate and what the reserve earns, for however long the decision is deferred. That is a real cost, paid in return for a real thing. It is an option, not a fund.

The named mistake, and it is common enough to be worth recognising: clearing the loan with the emergency reserve or with money earmarked for the first years of retirement, arriving at the stop date with a paid-off house and nothing liquid. A house cannot be sold in slices, and the buffer it replaced was the thing that stopped a market fall becoming a forced sale — the emergency fund does not become less necessary as the loan shrinks.

When funding other people stops being free

Refuse the assumption first. Some readers in their fifties are funding a child's education or wedding; some a parent's care; some a sibling, a partner's family, a dependent adult or a business that has not turned; some are funding a household of two and nobody else; and some are funding nobody at all. The arithmetic below is the same whoever is at the other end of it, and the last of those cases has its own problem, dealt with at the end.

The number that makes this a fifties question rather than a general one is the multiplier. On the illustrative case above, ₹1 lakh a year of spending that is permanent rather than temporary raises the corpus you have to reach by 21.5 times that, or ₹21.5 lakh. A transfer that was affordable out of income at forty-eight becomes a claim on capital the moment the income stops, and it is usually still running. The specific error is not generosity. It is budgeting an open-ended commitment as a temporary one for the seventh year running, which is the same ratchet lifestyle inflation describes, applied to somebody else's costs.

There is also an inversion in who should be carrying a bounded, datable claim. Someone being funded at twenty-three has roughly forty years of earning ahead of them; the person funding them may have four. Both may need the same ₹20 lakh, and only one of them can borrow against earnings not yet received. That is not a statement about who deserves the money — it is a statement about which of the two shortfalls has a lender in it. The full asymmetry, and the arithmetic of what a withdrawal from retirement savings costs in income terms, sits in the forties article and the education loan article.

The version of this that does real damage is the corpus used as the family's lender of last resort. By the fifties it is often the largest liquid pool anyone in the extended family has, which makes it the obvious place a request lands. Every other pool in that family has a replacement — time, income, a loan, a smaller version of the plan. The retirement corpus has no replacement, at any price, and that is the only reason it should be treated differently from any other savings.

Some claims are genuinely not discretionary. An older parent needing care, or an adult who will need support permanently, is not a decision to be optimised; pretending otherwise is not planning. What changes is where it goes: into the corpus target as a permanent line, at the 21.5 multiple, rather than into the residual category of what is left over each month. The timing mechanics of insuring against a health event, as opposed to funding one, are in the next section and in medical emergency planning.

Now the mirror case, which gets written about far less. A reader with no dependants is usually told they have the easy version of this decade. In one specific respect they have the harder one: where no household member would, by default, provide care, administration or a spare room in later life, those are purchased, not assumed, and they belong in the spending figure as an explicit line. That line is frequently larger than the transfers the other reader is making, and it is invisible because nothing leaves the account today. Who holds authority if you cannot act for yourself becomes a document question rather than a family one — power of attorney, nomination and a will.

Cover is priced on the body you have, not the one you had

Health cover is the one item in this article whose price is set by something money cannot reach: the health you have on the day the insurer assesses you. Every other decision here can be made better by having more money. This one cannot be made later.

Three clocks decide most outcomes, and all three run in calendar time, not rupees. The first is underwriting: an insurer prices and accepts the risk in front of it, so a condition that has already appeared is priced, excluded or declined rather than covered on ordinary terms. The second is the pre-existing disease rule — under the IRDAI product regulations a condition diagnosed or treated within 36 months before the policy is written is pre-existing, and it can carry a waiting period of up to 36 months. That clock starts when the policy starts, not when the condition did, which is why buying cover after a diagnosis does not pay for that diagnosis.

The third is the moratorium. After a policy has run 60 months the insurer's ability to contest a claim on grounds of non-disclosure narrows — and the clock runs afresh on any increase in the sum insured. What the moratorium does and does not protect in practice is a question for the policy wording, and the health insurance guide is where the product itself is explained.

None of those clocks can be started retrospectively, which is what puts them in this decade rather than the next. A wait begun while still working is served with income still arriving and, for anyone holding an employer policy, cover already in place behind it; the same wait begun after the stop date is served with neither, and whatever happens in the interval is paid out of the corpus. The waits are the same length either way — what changes is what is standing behind them.

Put those together and a trade-off appears that has no comfortable side. Medical costs rise while a sum insured does not, so cover bought at fifty-four is smaller in real terms at seventy-four. Buying the larger cover now means paying premium for years on protection you do not yet need. Topping it up later means the increase is underwritten on an older body and starts a fresh clock on the added portion. There is no version in which the cover keeps pace for free, and the choice between one policy and several — including whether a shared floater or separate policies suit a given household — changes the answer, which is a structural question.

One fear about buying cover this late is worth answering with the rule rather than with reassurance: that once you are old enough to claim on it, the premium can be raised without limit until you give the policy up. It cannot. For indemnity individual health cover on someone aged sixty or above, premium revision is limited to 10% per annum under the IRDAI circular of January 2025.

The sharpest trap in this decade belongs to people who have only ever held employer group cover. Group cover is the employer's contract, not yours, and it ends with the employment. A person insured that way for twenty-five years reaches their stop date with zero accrued waiting periods of their own, at the age when a fresh individual policy is slowest and hardest to write. Whether the group policy converts, on what terms, and whether accrued periods carry across is a question to put to the policy document and the insurer while still employed — not one to assume in either direction.

Two mechanics make an existing policy more portable than most holders think. Switching insurer at renewal carries accrued waiting periods across, provided the request lands at least 30 days before, and not earlier than 60 days from, the renewal due date — the whole subject of portability. And a new policy can be returned within 30 days of receiving the document. Cover against a condition that stops the earning rather than only costing money is a separate instrument, dealt with in critical illness and disability cover, and it matters here because a forced stop at fifty-six removes the highest-value earning years in the sequence.

The last decade in which the buffer can be built rather than sold

Somewhere in this decade the corpus stops being a single number and becomes a schedule. The question changes from how large it is to which part of it funds the first few years, and that question has a deadline attached that the number does not.

The reason is mechanical. Withdrawing during a fall sells more units to raise the same rupees, so the order in which returns arrive matters as much as their average — the full treatment, including what a withdrawal rate can and cannot tell you, is in safe withdrawal rates. The usual answer is to hold the first years of spending in something whose value on a stated date is not in question, which the bucket approach describes.

What is specific to the fifties is not the buffer. It is where the buffer comes from. While contributions are still arriving, the near-dated holdings can be built, not sold — new money directed there changes the shape of the portfolio without any disposal. After the last contribution, every top-up of that buffer and every rebalance back toward the target is a sale, and a sale in a falling market is precisely the event the buffer exists to prevent. The capability to reshape a portfolio without selling anything expires with the income, and it is the least-discussed thing the last working years are for. How the rebalancing itself works is in rebalancing.

Disposals also stop being free at this portfolio size. Long-term gains on equity and equity-oriented fund units are taxed under s.198 of the Income-tax Act 2025 at 12.5% above an annual ₹1.25 lakh, and that applies under either regime. Directing new contributions at whatever is underweight moves the ratio without triggering any of it, which is the same lever described above and another reason it is worth using while it exists.

One tax point catches retirees specifically and is routinely missed. Under s.156(2) the default new regime carries a rebate of up to ₹60,000 where total income does not exceed ₹12 lakh — but s.156(3) provides that the rebate cannot shelter special-rate income, and s.198 long-term capital gains are exactly that. Someone whose income after stopping work is mostly gains rather than salary or interest therefore does not get the shelter the headline threshold suggests. Which of the two regimes suits a given year is settled in the regime comparison, and how the income is assembled in the first place is retirement income sources.

What still moves the answer at fifty-six

The useful thing about a short horizon is that it makes the levers countable. There are four, they can be ranked, and the ranking is not the one that applied twenty years ago.

Take the illustrative case forward. A reader at fifty-six holds ₹1.2 crore, adds ₹9 lakh a year, and stops at sixty. By then the corpus is ₹1.73 crore — ₹1.2 crore grown four years at the illustrative real 3%, which is ₹1.35 crore, plus four contributions worth ₹37.65 lakh. Against the ₹1.93 crore the thirty-five-year spending needs, that is a gap of ₹20.7 lakh. Everything below closes the same ₹20.7 lakh, and the horizon stays anchored to age ninety-five, so working an extra year removes a year from the far end rather than pushing it outward.

The order is the point. At thirty-six the contribution lever dominates every other, because there are three hundred instalments for it to work on and the multiplier on spending is decades away. At fifty-six the same lever needs a 55% increase to do what ₹8,000 a month of lower planned spending does by itself. The levers do not weaken evenly as the horizon shortens; the ranking inverts. That is the single most useful consequence of the constraint this article is about, and it is invisible unless the two multipliers are written next to each other.

Two honesty notes. The real rate here is a round 3% for legibility rather than derived by dividing a nominal return by an inflation rate, and the corpus article quantifies what that shortcut costs. And the deferral lever — whose full arithmetic the forties article works through — is worth more here than it was there, because it acts on a larger balance and a shorter funding period, while being not a lever anyone controls. Health, sector conditions, employer decisions and caring responsibilities each remove it, usually at short notice. A plan resting on three extra years has an unhedged assumption in it.

What genuinely cannot be changed is worth naming too, because effort spent there is effort not spent on the four rows above: the years already elapsed, the returns the market has already delivered, and the medical history an insurer will read. Everything else in this decade is still a choice, and which assumption moves the answer most is where to start testing it.

Testing the plan against the record rather than the average

Every figure above runs on a constant real return, which is a modelling convenience and not a property of any portfolio. That gap matters more in this decade than in any earlier one: a plan that reaches its corpus along the average path may not reach it from the start dates that actually occurred, and there is no second attempt behind it.

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 contribution schedule on any scheme and period and reports the return on cashflows landing on irregular dates — XIRR — together 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 the arithmetic above: whether the four remaining years of contributions reach the corpus using the worst historical window rather than the mean, and how deep a fall the near-dated buffer would have had to absorb in the two years either side of a stop date.

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 what happens to the purchasing power of the income once it starts is the subject of inflation after you stop earning.

Common questions

Continue reading

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