Monte Carlo retirement planning: what a success rate does and doesn't tell you

One path versus many

A deterministic projection, the kind most calculators show first, picks one return and applies it every year. It's useful for seeing how accounts, taxes and benefits interact, but markets don't deliver steady returns. Some years are well above average and some are well below.

A Monte Carlo simulation, named after the casino district in Monaco because it relies on random draws, asks a different question: if returns vary from year to year around the same average, how often does the plan still work? Instead of one answer, you get a spread.

That spread matters because retirement involves withdrawals. A plan that works on average can still fail if poor years land at the wrong time, which a single-line projection can't show.

How the simulation works, in plain words

The planner runs your plan 500 times. In each run, every year's return is drawn at random around the average you entered, with the size of the swings set by a volatility input. A volatility of 10% means typical years land within roughly 10 percentage points of the average, with occasional years further out.

Everything else stays the same in every run: your contributions, retirement age, CPP and OAS, taxes, RRIF minimums and spending target. Only the order and size of returns change.

At the end, the planner counts the runs where spending was fully covered every year to the end age. That share is the success rate. It also sorts the account balances at each age to show percentile bands, described below.

Leah and Sam's plan on a straight line

Suppose Leah and Sam are both 55 and live in Nova Scotia. Together they have $420,000 in RRSPs, $110,000 in TFSAs and $30,000 in a non-registered account. They add $18,000 a year to RRSPs and $14,000 to TFSAs until they retire at 63. Their CPP estimates at 65 are $1,050 and $900 a month, and both expect full OAS. They'd like to spend $70,000 a year after tax, to age 95.

With illustrative returns of 6% before retirement and 5% after, and 2% inflation, the steady projection looks comfortable. Their savings reach about $1,064,370 at 63, the money never runs out, and about $423,389 is left after tax at 95. The calculator estimates their highest steady spending to 95 at $77,700 a year.

For couples, the planner assumes taxable income is shared evenly between the two partners, a simplification that pension income splitting can approach in practice. The couples page covers this in more depth.

The same plan across 500 market paths

Run through the simulation with 10% volatility, the same $70,000 plan succeeds in 59% of paths. About 5% of paths run short before 80, about 15% before 85, about 28% before 90 and about 41% before 95.

Nothing about the plan changed between the two views. The difference is entirely the variation in returns, which means some paths hit poor years early and never recover enough to support the spending target.

That's why a projection that looks safe on a straight line can deserve a second look. The steady result says the plan works if returns are average every year; the simulation says what happens when they aren't.

Reading the percentile bands

Alongside the success rate, the simulation shows bands. The 50th percentile (median) is the middle result at each age: half the paths are above it and half below. The 10th percentile is a weak outcome, worse than 90% of paths; the 90th percentile is a strong one.

For Leah and Sam at $70,000, the combined balance at 75 ranges from about $209,071 at the 10th percentile to about $622,658 at the median and about $1,402,342 at the 90th. By 85, the 10th percentile has reached zero while the median still holds about $355,571.

The width of the bands is information in itself. A wide spread means the outcome depends heavily on markets; a narrow one means the plan is less sensitive to them.

What a lower spending target does

Testing a lower target shows how much room spending choices create. At $62,000 a year, the same couple's success rate rises to 87%. Only about 7% of paths run short before 90 and about 13% before 95.

The weak outcomes improve the most. At 85, the 10th-percentile balance is about $135,121 instead of zero, and the median is about $670,960.

A difference of $8,000 a year in spending moves the success rate by 28 percentage points in this example. Whether that trade-off is worth it is a personal judgment, but the simulation makes it visible.

Limits worth keeping in mind

A simulation is only as realistic as its inputs, and it simplifies the world in ways worth knowing about:

• The average return and volatility are assumptions you choose, not forecasts. Different inputs give different success rates.
• Each year's return is drawn independently, so the simulation doesn't model long stretches of linked good or bad years.
• Spending is held steady in today's dollars; real retirees often adjust after a poor year, which tends to improve outcomes.
• A path that runs short at 94 counts as a failure just like one that runs short at 75, so the success rate alone hides how close the misses are.
• Health costs, home equity, inheritances and changes in tax rules aren't part of the simulation.
• Tax estimates exclude Quebec provincial income tax.

Using a simulation alongside other checks

A simulation works well as one view among several. The steady projection shows how accounts, taxes and benefits interact year by year; the simulation shows how sensitive that picture is to markets. Looking at both side by side gives a fuller sense of a plan than either alone.

The guide on sequence of returns risk explains why the timing of poor years drives most of the failed paths, and the 4% rule guide applies the same simulation to a common rule of thumb.

To try it with your own numbers, run the retirement plan tool and compare a few spending levels or volatility settings, then use the plan summary to see the key results together. The methodology page describes the simplifications the planner makes.