Retirement: What It Costs and Why It Starts Earlier Than People Think

Retirement costs one number and one duration: what the household will spend each month once the earned income stops, and how many years that has to run. Both are estimates and neither can be looked up. Retirement starts earlier than people expect because the arithmetic rewards time far more than amount, so five years of delay costs several times the contributions those five years would have carried.

Underneath that answer sits one asymmetry, and everything below is a consequence of it. Every other goal a household sets can be missed and then recovered from. A school fee that could not be paid this term gets paid next term out of next term's earnings. A roof that leaked through one monsoon gets mended after the next good month. The earnings are what make the recovery possible, and the earnings are still there. Retirement is the only goal where the earnings that would fund the recovery are the earnings that have stopped, so the arithmetic is worth doing badly and early rather than well and late.

What does retirement actually cost?

The word carries more baggage than it deserves. Start there. Retirement is not an event with a ceremony and a garland. Retirement is a date after which the earned income stops. A service record may have ended, a body may no longer manage the work, or nobody may be offering the work. A tailor at a counter in a market lane never retires in any formal sense: the eyes go, the hands slow, orders move elsewhere, and the counter takes less and then nothing. The event has happened and no letter was issued for it.

So the cost of retirement is not a lump the way the cost of a scooter is a lump. The cost is a monthly amount multiplied by a number of years. Retirement costs whatever the household would spend each month once the earned income stops, for however long that period runs, and both of those are estimates rather than facts.

Notice how unusual that is. Most household goals carry one estimate at most: a wedding two years out has a cost that can be asked about, and school fees are printed on a circular. Retirement carries an amount nobody has spent yet and a duration nobody can know, and it is the largest number the household will ever put on paper. Postponing it is an ordinary response to an uncomfortable sum rather than a character flaw.

Richard Thaler's work on how people weigh outcomes at different distances says the rest. A cost twenty four years out is discounted so steeply that it barely registers against a bill due on Friday, and the bill due on Friday is real. Discounting a distant cost is the ordinary shape of human attention rather than laziness, and it is why this arithmetic has to be written down: written arithmetic does not fade with distance the way an intention does.

How Retirement Planning Works, and what are the four inputs?

Strip away every scheme name and every product and the calculation has four inputs. Not five, not twelve, and no software is required to hold them.

The first is what it would cost a month once the income stops, in today's money. The second is how long that has to run. The third is what is already set aside for it, wherever it sits. The fourth is what is going in each month from here. Every retirement calculation ever performed is those four numbers arranged against each other, and the only output is a gap.

Three of the four can be estimated by a person sitting at a table with the household's own papers. The second cannot be looked up by anybody, at any price, by any method, and a calculation with an unknowable input has a range as its honest output rather than a single figure.

There is a fifth number that people mistake for an input, and it is worth naming now so it can be watched for later. The assumed rateA figure somebody chooses so the arithmetic can run. No rate becomes more likely than another because somebody typed it into a box. is the rate at which whatever is set aside is imagined to grow. The assumed rate is not an input the household holds. The rate is a choice made so the sum can be completed, and the further out the goal, the more of the printed answer is that choice rather than the money. An accumulation figure means little until the contributions behind it are set beside it.

What is a Retirement Expense, and why is it not this month's spending?

A retirement expenseWhat a household would spend in a month after the earned income has stopped. The later list is a different one from what the household spends now, not a smaller version of the same list. is what the household would spend in a month after the income stops. The trap sitting inside that sentence is the assumption that the later figure is this month's spending with a haircut applied. The later figure is a different list.

Take the Bhosale household, an invented household that runs through everything below. Meghna Bhosale is salaried at Sahyadri Freight Services Private Limited, also invented. Ashok Bhosale runs a tailoring counter in a market lane and what it takes changes every month. Ira Bhosale is seven and in school. At 31 March of year two, Rs 42,770/- leaves that household in an ordinary month: rent, maintenance, phone and broadband, groceries, electricity, gas, fuel, medicines, eating out, and one twelfth of everything that arrives once a year.

Now ask what a month looks like twenty four years later. Ira Bhosale is thirty one and running her own household. The school terms ended long ago, nobody commutes to a freight office, the life cover premium has finished its term, and two people eat instead of three. Against all of that, one line has grown: medicines, consultations and health cover, Rs 1,840/- a month between them now.

The retirement expense is not the current figure discounted; it is a fresh list of what a different household, in a different phase, would actually spend. Some things come off it, some things go on it, and one line changes shape entirely. The household's own estimate, done on one sheet at the kitchen table, is Rs 30,000/- a month in today's moneyPriced at what things cost now, so the figure can be compared with what the household spends now. The figure says nothing about what those things will be priced at later., Rs 3,60,000/- a year. The Rs 30,000/- is the household's own figure.

How Retirement Expenses Change Over Time, and what comes off the list?

Here is the sheet, line by line, so the fall from Rs 42,770/- to Rs 30,000/- can be checked rather than believed. Five things come off and one goes on.

Read the middle column before the right one. Not a single line in it is a resolution to spend less. The fall of Rs 12,770/- a month comes from the household changing shape, not from shopping more carefully. Two people in a smaller place with nobody at school and nobody commuting are a different unit from three people with a child, a job and a market counter, and cost a different amount to run.

Misreading the fall has a cost. Somebody who reads the later figure as the earlier one with discipline applied concludes the same discipline is available today, and nobody can decide to stop having a child in school. Phases end on their own schedule.

Why does the total fall while the health share rises?

One line across the two sheets shows the whole shape of later spending. Medicines and health cover together are Rs 1,840/- a month out of Rs 42,770/- now, or 4.3 per cent of the sheet. Later the same line is Rs 5,180/- out of Rs 30,000/-, or 17.3 per cent.

The total falls by about thirty per cent while the health line nearly triples and its share of the sheet quadruples, and those two movements at once are what makes a retirement expense hard to estimate by feel. A halved grocery bill is noticed immediately. Health spending arrives one consultation at a time across a decade, so a quadrupled share is noticed only in retrospect.

The health line also behaves differently from every other line on the sheet. The rest of the sheet is decisions the household makes month by month, and a bad month is absorbed by making a smaller decision. The health line arrives without asking, at a size nobody chose, and more often as the years go. A grocery bill can be trimmed in a bad month; a hospital admission cannot be postponed until a better one.

None of that tells anybody what to do about it. The arithmetic says only that an estimate built by cutting today's number gets the total roughly right and the composition badly wrong, and the composition decides whether the figure holds up when it is tested.

How long does the money have to last?

Duration is the second input, and the one nobody can settle. The durationHow many years the money has to cover, counted from the day the earned income stops. Nobody can know the duration in advance, so the honest output of the calculation is a range. is the number of years between the income stopping and the household no longer needing the money. Asked plainly, the difficulty is obvious: how long will a person live? Nobody has an answer, and published averages describe populations rather than the person reading this sentence.

So do the only honest thing available. Run the same multiplication at several lengths and look at the width. At Rs 30,000/- a month in today's money, twenty years needs Rs 72,00,000/-. Twenty five years needs Rs 90,00,000/-. Thirty years needs Rs 1,08,00,000/-. Nothing has changed between those three lines except the number of years, and the answer has moved by Rs 36,00,000/-.

One input nobody can supply moves the requirement by thirty six lakh rupees, more than twice everything this household holds today, and no care with the other three closes that width. The width is not a reason to skip the sum. The width is why the sum produces a range, and why a single confident figure for what a household needs has quietly resolved something nobody can resolve.

Two things are deliberately left out of that multiplication. Prices rising are left out, so the figures stay in today's money, and so is any return on what has not been spent yet. Both matter and both are treated elsewhere. Both are left out so that one thing stays visible at a time, and that one thing is the duration on its own.

What do prices rising do to a figure quoted in today's money?

The Rs 30,000/- is priced at what things cost now, and twenty four years is a long time for prices to sit still. Nobody knows at what rate they move, and the official series belongs elsewhere. The arithmetic below shows what choosing between two assumptions does to the answer.

Take the same Rs 30,000/- basket and price it twenty four years later on three assumed rates, all three labelled assumptions and none of them described as typical or expected. At an assumed 4 per cent a year the basket is Rs 76,899/- a month. At an assumed 6 per cent it is Rs 1,21,468/-. At an assumed 8 per cent it is Rs 1,90,235/-.

The distance between the lowest and the highest is Rs 1,13,336/- a month, more than two and a half times everything the household spends in total today, and it is produced entirely by which assumption somebody typed in. Sit with that. The household built the Rs 30,000/- line by line from its own sheet and controls every rupee of it. The household controls nothing at all about the rate that turns Rs 30,000/- into one of those three, and that rate moves the answer further than the estimate does.

The Reserve Bank of India at rbi.org.in publishes the official price series for this country, and a figure for what prices have actually done belongs to that source. The arithmetic on its own carries the point: over a long enough goal, the assumption becomes larger than the goal.

How Delaying a Retirement Saving Start Changes Outcomes, and what does the arithmetic say?

Now the counter-intuitive part, and the reason the title says what it says. Everything so far was the size of the goal. The putting in begins here.

Meghna Bhosale is 36 at 31 March of year two and would reach 60 in 24 years, or 288 months. Rs 6,240/- a month goes into her provident fundA workplace arrangement that accumulates money put in from the pay side and from the employer side. The provident fund is taught in full separately; here it is only the place the monthly amount goes., being Rs 3,120/- deducted from her pay and Rs 3,120/- added by her employer. The rates behind those two amounts are set by scheme rules and change, so both are stated as flat rupee amounts rather than computed from a rate.

Run that Rs 6,240/- a month for 288 months on an assumed 8 per cent a year, a rate chosen so the arithmetic can be shown rather than a forecast of anything, and it reaches Rs 54,07,867/-. Of that, Rs 17,97,120/- is the money itself, being Rs 6,240/- multiplied by 288. Now delay the start by five years. The contributionsThe money actually handed over, month by month. A contribution needs no assumption at all, unlike whatever the arithmetic adds to it. are unchanged, so the only thing that has moved is the starting date. The run is 228 months and reaches Rs 33,22,070/-.

Removing five years from the front of a twenty four year run removes Rs 20,85,797/-, against Rs 3,74,400/- of contributions that were never made. The delay costThe difference between what a run reaches starting now and what the same run reaches starting later. The contributions never change; only the starting date does. is mostly time: the money not put in is a fifth of what was lost.

Here is the sentence worth reading twice. The years a late start removes are at the front of the run, but the slice of the curve that vanishes is at the top: a person who starts five years late does not end up missing the cheap flat beginning, but with a shorter curve that never reaches its steep part. Look at the shape below. The first ten years add Rs 11,41,583/- and the last five add Rs 20,85,797/-. Each early year is the one with the longest run still ahead of it.

The three settings the slider passes through are worth having in words as well. Starting now, 288 months, reaches Rs 54,07,867/- against Rs 17,97,120/- put in. Starting five years late, 228 months, reaches Rs 33,22,070/- against Rs 14,22,720/- put in. Starting ten years late, 168 months, reaches Rs 19,22,061/- against Rs 10,48,320/- put in. Five years of delay therefore costs Rs 20,85,797/- against Rs 3,74,400/- skipped, or 5.6 times what was not paid in.

One thing worth doing before reading on. Drag the slider to one year: Rs 4,86,185/- lost against Rs 74,880/- never paid in, a ratio of 6.5 times. Then drag it to fifteen and the same ratio has fallen to 3.9. The very first year of putting it off has the longest run still ahead of it, so it is the most expensive single year of the whole delay. A first year of delay does not feel that way at all.

Why does the delay cost several times what was skipped?

The ratio is doing the teaching here rather than the figure. Rs 20,85,797/- depends completely on the assumed rate, so a different rate changes it. The ratio of what the delay cost to what it skipped is far steadier, and it survives changing the assumption.

Think about a fruit tree rather than a bank. A tree planted five years earlier is not five years taller; it is taller by five years of growth applied to everything it had already grown. The gap between two trees planted five years apart therefore widens every season instead of holding still. Money left where it adds to itself behaves the same way.

A delay costs a multiple of what it skipped because a contribution buys two things at once, the money and the years the money then sits, and only the first can be bought back later. Somebody who skips Rs 6,240/- this month can put in Rs 12,480/- next month and be square on the money. Nothing restores the month of sitting.

What does the whole calculation look like for one household?

Put the four inputs and the delay arithmetic on one sheet for the Bhosale household. Every figure below belongs to that household and to no other.

Two things about that sheet deserve saying out loud. The first is the size of the shortfall. At Rs 30,000/- a month for twenty five years the requirement in today's money is Rs 90,00,000/-, and the run to 60 reaches Rs 54,07,867/- before prices rising are considered at all. The gap is large, it is arithmetic rather than a judgement, and seeing it early is the only advantage available.

The second is how much of the Rs 54,07,867/- is money. Rs 17,97,120/- was handed over across 288 months, a third of the total; the other two thirds is what the assumed rate adds and is not money anybody holds. Watch that proportion as the run shortens: at 228 months the money is 43 per cent of the answer, and at 168 months it is 55 per cent.

Long periods are usually presented the other way round. A longer run is a less certain calculation rather than a more certain one, since more of its answer rests on a rate somebody chose and less on contributions somebody controls. An accumulation figure is worth reading only beside the contributions that produced it.

What does the ratio mean for a household that is already late?

For a reader arriving at fifty, or fifty five, with nothing set aside and no scheme behind them, the arithmetic above may read like a bill for a decision they never got to make. The bill is not owed, and the arithmetic itself says why.

Go back to the shape of the answer. With 288 months to run, a third of the figure is money put in and two thirds is what the assumed rate adds. With 168 months, 55 per cent is money. With 120 months, about two thirds of it is. The shorter the run, the less the answer leans on an assumption nobody controls and the more it leans on the contributions. The contributions are the one part of this calculation anybody holds.

A late start is a genuinely different calculation rather than a consolation prize dressed as arithmetic. There is less time for the time to work, so somebody starting at fifty is running a sum in which the amount matters more than the time. The four inputs are unchanged; what has changed is which carries the weight, and it is now the one that answers to a decision rather than to a rate somebody typed into a box.

The duration input works the same way, and the blunt version is more use than the polite one. A household whose earned income stops later is running a shorter period, and a shorter period needs less money. The shorter period is arithmetic rather than encouragement, and whether anybody should work longer is a different question. The four inputs go on working for somebody who arrives late.

One last thing, and it matters more than the sums. The commonest reason a household has nothing set aside is not that it decided badly. The reason is that there was nothing to set aside, month after month, for ordinary reasons: one income, a parent to support, an illness, a business that never quite took. Arithmetic is not entitled to an opinion about any of that.

Why is this the one goal a household cannot recover from?

Every other goal has a second attempt built into it, funded from the same place as the first. Ira Bhosale's school fee that could not be paid in April gets paid in May out of May's takings. A wedding the household could not afford in the year it was planned happens a year later on a smaller scale. In each of those the earned income repairs the miss, and the earned income is still arriving.

Retirement is the one goal where the miss and the loss of the repair mechanism are the same event, because the income that would fund the catching up is what has stopped. There is nothing to catch up out of. The difference is one of kind rather than size, and it is why retirement earns a place on paper twenty four years early even though every other bill on the table is more urgent.

The word recoverableWhether a missed goal can be made good later out of future earnings. Almost every household goal is. Retirement is not: the earnings are what stopped. is doing real work there. Almost everything else in household finance is recoverable, and the language of household money is full of second chances: consolidate, restructure, reschedule, start again next year. None of those verbs attaches to a retirement that was missed. The moves that remain are smaller and harder: spend less than the sheet said, lean on people who have their own sheets, or keep working past the point the body wanted to stop.

Saying that plainly is not a warning. Naming it is the one argument for putting an uncomfortable sum on paper long before there is any room to act on it.

What do people actually do with these two numbers?

Three uses, none of which requires an adviser, a spreadsheet or a scheme.

The first use is turning a vague weight into something that can be argued with. Before the sheet exists, retirement sits in a household as a feeling with no edges, and there is nothing in it to disagree about. Afterwards there are four numbers, and any of them can be challenged: whether Rs 30,000/- is too low, whether twenty five years is the right length to test, whether the health line is understated. An estimate that can be argued about is worth far more than a worry that cannot, even when the estimate turns out to be wrong.

The second use is fixing the order of decisions before the moment arrives. The sheet gets written, the ratio gets seen, and the household knows in advance what it does with the first month that has room in it. Deciding at a calm table beats deciding on the day the money appears, when everything else is competing for it.

The third use applies to somebody with no payslip at all. Ashok Bhosale's counter produces no statement, no employer side and no service record, and he can still write all four inputs in a notebook: what a month would cost, how long to test it for, what is already there, what is going in. The arithmetic does not check who employs a person, and every mechanism above works with no employer anywhere in it.

What can no arithmetic settle, however carefully it is done?

Two inputs, and every figure above rests on both. Naming them is the last piece of the mechanism rather than a disclaimer at the bottom.

The first is how long the money must last, and it is unknowable in the strict sense: not hard to estimate, not short of better data, unknowable. Every figure in the duration table moved by lakhs when that one input moved. The honest treatment is to run the sum at several lengths and read the width.

The second is the assumed rate. Every accumulation figure here, including the Rs 54,07,867/-, the Rs 20,85,797/- and the ratio of 5.6, comes from the same assumed 8 per cent a year. Nobody promises it, nobody can, and no probability attaches to it. Every number above is exact arithmetic resting on two figures nobody can supply, and that is a property of the subject rather than a weakness of the arithmetic.

The shape is what survives, and four statements carry it. The cost is a monthly amount times a duration. The later sheet is a different list rather than a smaller one. Time does more work than amount, so the front years are the expensive ones. And a figure shown without its contributions is hiding how much of itself was ever money. All four hold whatever rate anybody assumes and whatever duration turns out to be true.

The Bhosale household, Meghna Bhosale, Ashok Bhosale, Ira Bhosale and Sahyadri Freight Services Private Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.