When you withdraw while investments fluctuate, the order of returns matters. An average annual return alone cannot show whether a retirement income plan survives.
State the timing before comparing paths
Start with a hypothetical MYR100,000 portfolio. Withdraw MYR10,000 at the beginning of each year, then apply that year’s investment return to what remains. No new contributions, costs, taxes, dividends paid separately or inflation adjustments are included.
Path A returns +20% then −20%. Path B returns −20% then +20%. These are invented two-year paths, not a market forecast or a chosen historical backtest. Changing withdrawal timing would change the figures, so the timing is part of the example.
First remove withdrawals as a control
Without withdrawals, both paths finish at MYR100,000 × 1.2 × 0.8 = MYR96,000. Multiplication can be reversed without changing the result in this two-return example. The arithmetic average return is zero, but the compounded outcome is a 4% loss.
That difference already shows why averaging percentage changes is not the same as growing money. Now add the same MYR10,000 beginning-of-year withdrawal to both paths and the order can no longer be ignored.
Calculate the rising-first path
Path A first withdraws MYR10,000, leaving MYR90,000. The +20% return makes the year-end balance MYR108,000. Year two starts with another MYR10,000 withdrawal, leaving MYR98,000; the −20% return then leaves MYR78,400.
The total withdrawn is MYR20,000. It should not be subtracted again from the ending balance. Money already spent is part of the cashflow history, while MYR78,400 is the remaining investment balance.
Background source: Fidelity: retirement risks, including return sequence (background only)
| Path | End year 1 | End year 2 |
|---|---|---|
| +20%, −20%; no withdrawals | MYR120,000 | MYR96,000 |
| −20%, +20%; no withdrawals | MYR80,000 | MYR96,000 |
| +20%, −20%; MYR10,000 each year start | MYR108,000 | MYR78,400 |
| −20%, +20%; MYR10,000 each year start | MYR72,000 | MYR74,400 |
Calculate the falling-first path
Path B first withdraws MYR10,000 from MYR100,000, leaving MYR90,000. A −20% return leaves MYR72,000. After the second MYR10,000 withdrawal, only MYR62,000 participates in the +20% year, ending at MYR74,400.
Both paths received the same two returns and paid the same MYR20,000 total withdrawals, but Path B ends MYR4,000 lower. The early withdrawal after a loss leaves less invested for the recovery. A later rise does not act on the original portfolio size.
Explore responses without promising a cure
Possible responses include a spending adjustment, other dependable income, accessible reserves or a portfolio suited to the time horizon. Each has trade-offs. Holding more cash can reduce forced selling but can also lower expected growth and expose purchasing power to inflation.
A reserve is part of the household’s assets, not extra wealth added after the simulation. If spending is inflexible, saying you will cut it after a fall is not a credible plan. Identify discretionary items before relying on a flexible withdrawal strategy.
Know what a constant-return calculator leaves out
A smooth projection can compare inputs, but this two-year exercise shows a path effect it does not measure. It says nothing about a 30-year success probability, the correct asset mix or an appropriate personal withdrawal rate. Larger studies also depend on the model, data and spending rules chosen.
Use the manual sequence file to reproduce these two paths. Financial planning on Numbrite remains a deterministic tool; it does not run a retirement sequence simulation. Read the flexible-spending lesson next to compare what changes when the withdrawal itself changes.
Download the worksheet
Put it into practice
Check both paths manually before comparing withdrawal rules.
Compare spending adjustmentsContinue the financial independence lessons
Found an error? Contact NumNum
Put this lesson into a decision
These are separate local scenarios. Loading or saving one does not change your other Numbrite records; assumptions must match the case you want to test.