SWP Calculator

A systematic withdrawal plan takes a fixed amount out of an invested pot at regular intervals while whatever remains keeps earning. Two forces run against each other: growth pushes the balance up, withdrawals pull it down, and the outcome depends on which is larger at the size the pot happens to be. This page simulates that month by month rather than applying a formula, because once withdrawals enter the picture there is no closed form that stays honest across a changing balance.

The order inside each period is worth stating plainly. The withdrawal comes out at the start, then the remaining balance grows. That is the conservative reading and it matches how someone living off a portfolio experiences it, since the rent falls due before the quarter's returns arrive. It also means the pot never earns on money you have already spent.

Why the balance stops at zero rather than going negative

If a withdrawal is larger than the balance available, the simulation pays out what is there and stops. It does not carry a negative balance forward. A negative pot would keep compounding into a larger negative number and produce a tidy looking figure that describes nothing, because in reality the account is empty and the payments stop. So the final month of a failing plan often pays less than the amount requested, and the total withdrawn reflects what the corpus could actually deliver.

This is why the money lasts figure is the one to read first. An ending balance of zero tells you the plan failed but not when, and a plan that empties in year eight and one that empties in the final month both show zero at the end. The duration separates them, so it sits in the results panel rather than being left for you to infer from the table.

The delayed start, and what it is worth

Withdrawals can begin later than month zero, which is the common case: you retire on a particular date, but the pot has already been invested for some time. Returns compound from month zero regardless, so the delay is not dead time. On the page defaults the two year delay does real work. Drawing 400 a month from 50,000 at 7% starting immediately, the corpus lasts 18 years 6 months. Waiting two years takes it to 27 years 9 months. Two years of patience bought 9 years 3 months of income.

There is a withdrawal amount at which growth exactly replaces what you take out and the balance never moves. Above it the pot shrinks and eventually empties. Below it the pot grows and the plan runs indefinitely. At 50,000 earning 7% compounded monthly that figure is 289.98 a month, or 6.9594% of the corpus a year. Everything else about a withdrawal plan is a consequence of where your number sits relative to that one.

The arithmetic is short. Write F for one period of growth, so a balance left alone becomes itself times F. Because the withdrawal comes out first, the balance has to climb back from a smaller base, and the amount that holds it exactly flat is the balance multiplied by one less than F, then divided by F again. Skipping that final division gives the end of period answer, 291.67 rather than 289.98, which overstates the safe figure by about six tenths of a percent. The page default of 400 a month is 9.6% of the corpus a year, which sits 37.9% above the ceiling, and that is the whole reason the default plan drains.

Duration collapses far faster than the withdrawal rises

Most people expect something roughly proportional, that taking 20% more empties the pot around 20% sooner. It does not work like that. Holding the corpus at 50,000, the return at 7% and withdrawals starting immediately: 300 a month lasts 48 years 9 months, 350 lasts 25 years 4 months, 400 lasts 18 years 6 months, 450 lasts 14 years 10 months and 500 lasts 12 years 6 months. Going from 300 to 350, a rise of 17%, cost more than twenty three years.

The practical consequence is that precision matters most when you are close to the ceiling and barely matters when you are far above it. Drawing 500 a month from 50,000, the difference between assuming 6% and assuming 8% is a few years either way. Drawing 300, that same gap is the difference between a plan that lasts half a century and one that never ends.

Frequency is not free, and yearly is not twelve times monthly

Taking 4,800 once a year is not the same as taking 400 a month, even though the annual total is identical. The yearly version pulls the full amount out at the start of the year, so that money spends the following eleven months not earning. From 50,000 at 7% with no delay, monthly withdrawals last 18 years 6 months and the equivalent yearly withdrawals last 17 years 1 month. Seventeen months of income, given up to timing alone.

The same effect moves the ceiling. At 50,000 and 7% the sustainable monthly figure is 289.98, but the sustainable yearly figure is 3,370.83, not the 3,479.70 you get by multiplying by twelve. That extra 108.87 a year looks like a rounding difference and empties the corpus after 49 years 1 month.

This calculator applies one constant return for the whole term. That is a reasonable way to compare scenarios and a poor way to predict an outcome, and for a withdrawal plan the gap between those two things is wider than for any other calculator on this site.

Sequence of returns, which hits a withdrawal plan hardest of all

Take ten annual returns: 25, 20, 15, 12, 10, 8, 5, 2, 0 and 3 percent. Their arithmetic mean is exactly 10% and their compounded average is 9.7350%. Run 50,000 through them while withdrawing 400 a month. With the strong years first the corpus finishes at 60,805.31. With the identical returns reversed, strong years last, it finishes at 30,753.13. Same numbers, same average by either definition, 30,052.18 apart. The good ordering ends with 97.7% more money than the bad one.

Note the direction, because it is the reverse of the accumulation case. When you are paying into a plan, weak early years help, since they let you buy while you own little. When you are drawing down, weak early years are what ruins the plan, because you are selling into them at the point the pot is largest. A saver and a retiree looking at the same market are not exposed to the same risk, and not in the same direction.

Push the withdrawal higher and the sequence stops changing the ending balance and starts deciding whether there is an ending at all. At 600 a month the favourable order finishes with 27,909.92 still invested and the unfavourable one is empty after 8 years 2 months. At 800 both fail, but one fails after 9 years 6 months and the other after 5 years 8 months, a difference of 3 years 10 months. For a lump sum left alone this experiment produces nothing whatsoever, because multiplication commutes and reordering the factors cannot change the product. Order only bites when the amount at risk is moving, which is exactly what a withdrawal plan does.

Inflation, which turns a flat withdrawal into a shrinking one

A flat 400 a month is not flat in the sense that matters. At 3% inflation it has the purchasing power of 297.64 after ten years, 231.51 after eighteen and a half, and 221.47 after twenty. A plan paying a constant amount is a plan paying you less every year, and part of its apparent success rests on that erosion. The annual increase field corrects it: set it to your inflation assumption and the withdrawal holds its real value.

Doing so is expensive, and the size of the bill is the point. From 50,000 at 7% with no delay, a flat 400 lasts 18 years 6 months. The same 400 rising 3% a year lasts 13 years 9 months. Four years and nine months of income, and the only thing that changed is that the plan now holds its value instead of fading. The version that looked better was not better, it was quietly paying you less.

The ceiling moves too, and by more than people expect. A flat withdrawal from 50,000 at 7% can run at 289.98 a month indefinitely. One rising 3% a year can only run at about 169.64, which is 4.07% of the corpus a year. That figure landing near the familiar four percent guideline is not a coincidence, since that rule was built around an inflation adjusted income rather than a flat one. But treat the resemblance carefully: the rule exists because of the sequence risk described above, and this page assumes that risk away.