Why Inflation and Longevity Dominate Retirement Estimates

Lakshmi's plan was tested against the wrong thing

Lakshmi retired with a plan that had been checked. Somebody ran it at a lower return to see whether it still worked, it did, and that was treated as the stress test.

The return is the input everybody varies, because it is the one that feels financial. It is not the one that dominates. A retirement is an amount of spending, repeated for an unknown number of years, with the amount itself rising over that time — and both of those quantities are more uncertain and more consequential than the return assumption they sit alongside.

A plan tested only against a lower return has been tested against the least of its three uncertainties.

Inflation is not a background detail here

During a working life, inflation is partly offset by an income that also rises. In retirement that offset disappears. The spending keeps growing and there is no salary growing with it, so every year of inflation is a permanent increase in what the portfolio must deliver, compounded across whatever remains of a life.

The size of the uncertainty is worth seeing rather than assuming. Across the Indian consumer price record we hold — 151 months of year-on-year readings — the middle of the record sits at 4.9%, but 24.5% of those months were at six per cent or above and 30.5% were below four.

That is the range Lakshmi's plan has to survive, and a plan built on the middle of it is a plan built on a coin landing on its edge. The full spread is here, and the important point for a retiree is that the difference between the top and bottom of that range, compounded over decades, changes the required corpus by far more than a percentage point of return would.

There is a second problem underneath the first. Lakshmi does not buy the national basket. Her spending is weighted towards the categories that behave least like the average — medical care above all, and assistance of various kinds as she gets older. We hold no separate index for those, so this page cannot tell her how much faster they have risen, only that assuming they track the headline is an assumption rather than a measurement.

Spending does not follow a smooth line

The other reason to distrust a single inflation number is that retirement spending is not one thing growing at one rate.

Some categories genuinely fall with age. Travel, entertainment and the costs of going out to work tend to reduce, and a plan that inflates the entire budget mechanically for thirty years will overstate what is needed in the middle stretch.

Others rise, and one rises steeply. Medical costs and the cost of help with daily living are concentrated late, they arrive suddenly rather than gradually, and they are the largest single source of a retirement plan failing after it appeared to be working. A model that spreads a smooth inflation rate across a flat budget captures neither shape.

The honest version is to split the budget into a few categories with different paths and to plan for a late-life step up rather than a gentle slope. That is more work and it produces a more useful answer, because the failure it is protecting against is specific.

The number of years is a range, not a figure

The second dominant uncertainty is longevity, and it is uncomfortable in a way that makes it get handled badly.

Planning to a life expectancy is planning to roughly a coin flip. Half of people in any cohort live longer than its life expectancy, and the planning question is not how long Lakshmi will probably live but how long she might. Those have different answers, and only the second one is safe to build on.

Several refinements matter more than they appear to. Where there are two people, the relevant horizon is the second death rather than the first, which is materially longer than either individual expectation. Spending does not halve when one spouse dies — housing and most fixed costs continue — while some income may stop entirely, so the survivor's position needs its own line in the plan rather than being assumed to scale down. And late-life care is both the most expensive item and the one most correlated with living a long time, so the long-life case is not simply the central case extended; it is more expensive per year as well.

Why a higher return does not fix either

There is a tempting response to all this, and it is worth closing off.

If the corpus looks insufficient against high inflation and a long life, the arithmetic can be made to work by assuming a higher return. This changes the spreadsheet and nothing else, and it is worse than useless here because it introduces exactly the risk a retiree is least able to carry: a higher expected return means more volatility, and volatility early in retirement is the mechanism that does permanent damage.

So the response to a shortfall makes the plan more fragile at the point of maximum fragility. The levers that genuinely work are duller and all of them are within Lakshmi's control: saving more before retiring, retiring later, spending less, spending flexibly, and transferring some of the longevity risk to something that pays for as long as she lives.

Where risk-sharing fits

That last lever deserves a sentence of precision, because it is the only instrument that addresses longevity directly.

An income that continues for life, however long that turns out to be, converts an unknown number of years into a known monthly amount. That is what it buys — not a return, and it should not be assessed as though it were one. Against that, it gives up flexibility and whatever would have been left over.

The sensible shape is partial: enough guaranteed income to cover the spending that genuinely cannot be cut, with the portfolio funding the rest, where a bad year can be met by spending less. That is the flooring argument, and inflation is the question to ask of any such product — an income fixed in rupees for thirty years is a very different thing from one that rises.

What this page cannot tell you

Two gaps, both material, both stated rather than glossed.

We cannot show a drawdown running through a full retirement in real terms. The Indian equity record is shorter than a retirement and the inflation series shorter still, which means there is no complete inflation-adjusted period of the required length to test — none at all. That is set out in the limits of a safe withdrawal rate.

And we hold no Indian medical-cost inflation series, which is the single figure that would most change a retirement plan. The article can tell Lakshmi that her largest late-life cost probably runs faster than the headline; it cannot tell her how much faster, and it will not guess.

What to take away

Stress the plan against higher inflation and a longer life, not just a lower return. Those two compound against each other — more years of spending, each year costing more — and between them they move the required corpus far more than the return assumption does.

Split the budget rather than inflating it as one line, and plan for a late-life step up in care costs rather than a smooth slope. Use a horizon well beyond life expectancy, take the second death rather than the first where there are two people, and give the survivor their own line. Do not close a shortfall by assuming a higher return, because that adds volatility exactly where volatility is most damaging. And consider covering the uncuttable spending with income that lasts as long as you do, asking specifically whether it rises with prices.