How a Retirement Corpus Is Calculated

Kavita has been given a number with no working

Kavita has been earning well for years without a plan behind it, and she has recently started asking what she is supposed to be aiming at. The answers she has received are all the same shape: some multiple of her annual expenses, delivered as a fact.

She would like to know where the multiple comes from, and she is right to ask, because the honest answer is that it comes from a set of assumptions somebody else made — about how long she will live, how fast prices will rise, what her portfolio will earn, and what she will spend. Change any of those and the multiple changes. A number quoted without them is not a calculation; it is somebody else's calculation with the inputs removed.

A corpus is not a multiple. It is the amount that, on the day the salary stops, is sufficient to meet a schedule of future payments. Everything below is how to build that schedule.

The calculation, in five steps

Estimate the first year of retirement spending, in today's money. Not current spending — the spending of a retired version of the household, which is different. Commuting and work costs go. Children may or may not still be dependent. The mortgage may have ended. Medical costs are higher and rising. Getting this first figure roughly right matters more than any refinement that follows, because everything downstream is a multiple of it.

Split it into parts that behave differently. Some of the budget will fall as Kavita ages, some will rise steeply and late, and some will track the general price level. Treating the whole thing as one line growing at one rate is the most common simplification and it misstates both the total and its shape, for reasons set out in why inflation and longevity dominate.

Add the things that are not annual. A vehicle replaced twice. A roof. A family obligation. Late life care, which is the largest and the one most often missing. These are dated one-off amounts, not part of the recurring budget, and they belong in the schedule on their own dates.

Subtract dependable income, year by year. A pension, rent, anything that arrives regardless of the portfolio. This is a subtraction per year rather than a lump sum, because such income often begins at a particular age, and it changes on a death. What remains after the subtraction is the part the corpus actually has to fund — and it is usually much smaller than the gross spending, which is why this step is worth doing carefully.

Then bring it back to a value at retirement, using a return assumption for the money that will still be invested while it waits to be spent. This is where the schedule becomes a single number, and it is the step that hides the most.

The two mistakes that make the answer wrong

The first is a units error and it is silent.

Either Kavita inflates all the future spending into future rupees and discounts at a nominal return, or she keeps everything in today's purchasing power and uses a real return — a return net of inflation. Both are correct. Mixing them is not, and the result looks entirely plausible either way. Inflating the spending and then discounting at a real rate overstates the corpus badly; the reverse understates it. The check is to ask whether the number on the screen is rupees she could spend today or rupees she will spend in 2050, and confirm every rate on the page is in the same world.

The second is timing. Withdrawals taken at the start of each year require more capital than withdrawals at the end, because the money is not there to earn during the year. Over a long retirement the difference is not trivial, and two calculators using different conventions will disagree for that reason alone. The conservative convention — and the one a person actually living on the money experiences — is to take it at the start.

Fees and tax belong here too. Both reduce what a given portfolio can deliver, and a corpus computed on gross returns is a corpus that will fall short by roughly the amount that was omitted.

Why the answer is a range

The three biggest inputs are unknowable, and pretending otherwise is what produces the false precision Kavita has been offered.

She does not know how long she will live. She does not know what inflation will do over three decades. She does not know what her portfolio will return, and — this is the part usually missed — even if she knew the average return exactly, the order in which it arrived would change the outcome substantially, which is demonstrated here rather than asserted.

So the output should be several numbers. A central case, and then the stresses: higher inflation, longer life, lower return, and a poor first few years. Each of those tells her something specific. If the corpus barely moves under a stress, that input does not matter much and she can stop worrying about it. If it moves a great deal, she has identified what the plan actually depends on, and that is what to monitor.

What the number is for

The point of the exercise is not the number. It is the decisions the number informs, and there are only four of them.

How much to contribute. When to stop working. What to spend once stopped. And how much risk the portfolio should carry. A corpus estimate that does not change one of those has not earned the afternoon it took.

For Kavita specifically, the calculation is likely to produce an uncomfortable result, because starting in her forties is genuinely less forgiving than starting earlier — and she has said she would rather know. The useful response is to read the stresses rather than the headline: which of the four levers moves the plan furthest, and how much of the gap can each close. What to do when you cannot save enough takes that further.

Recalculate it, because it will be wrong

The last property of a corpus estimate is that it has a short shelf life.

It rests on assumptions that will be replaced by facts as time passes: actual inflation, actual returns, actual spending, an actual retirement date. Each annual review substitutes a little knowledge for a little assumption, and the estimate converges on something true.

An estimate reviewed every year is a planning tool. The same estimate made once and never revisited is a thirty-year forecast, and nobody should be running their retirement on one of those.

What to take away

Build the corpus from dated cash flows, not from a multiple: first-year retirement spending split into parts that behave differently, plus the one-off costs including late-life care, minus dependable income year by year, brought back to a value at the retirement date.

Keep nominal and real quantities consistent, decide whether withdrawals happen at the start or end of the year, and include fees and tax. Produce a range with stresses on inflation, longevity, return and a bad first few years, and read the stresses to find out what the plan actually depends on. Then recalculate annually, and treat any precise target that will not show its assumptions as somebody else's arithmetic.