Sensitivity testing in thirty seconds

A plan has a dozen inputs and they are not equally important. Finding out which ones your answer is genuinely sensitive to takes about half a minute, and it usually reorders what you assumed was worth worrying about.

The method

Move one input by a plausible amount, note how far the outcome moves, put it back. Repeat. Rank the inputs by the size of the move. That ranking is your plan’s sensitivity profile, and it is specific to you — a lever that dominates one household barely registers in another.

What usually dominates

For most households the ranking comes out in a similar order, and it is not the order people expect:

  1. Retirement age, because it moves contributions and withdrawals at the same time — the double effect from episode 69.
  2. The return you assume, and the fees you pay, which are one lever: a one percent fee costs exactly what a return one point lower does. Over a long horizon a point is worth more than people expect.
  3. Drawdown order, which is free to change and, for a household with a large registered balance, worth real money.
  4. Spending, which scales almost everything downstream of it.
  5. Benefit timing, which is usually worth something and rarely decisive.

The ordering is the point, and it is Priya’s: it moves with the household. What stays true is that CPP timing attracts most of the debate and, for her, moves the answer least.

Priya’s ranking

Priya — 54, Ontario, single. $118,000 salary, $410,000 in her RRSP and $88,000 in her TFSA, planning to retire at 63.

Against a baseline of $360,723 (which carries no fee at all): retiring three years early gives $40,618, a return one point lower gives $119,103, changing the drawdown order gives $556,490, spending ten percent more gives $195,082, and deferring benefits to 70 gives $396,549.

From Priya's $360,723 baseline, retiring three years early moves her most, to $40,618, then a return one point lower at $119,103, the drawdown order at $556,490, spending ten percent more at $195,082, and benefit timing least, at $396,549.
Five plausible changes, ranked by how far they move the answer.

Two things stand out. A single point of return — or, the same thing, a one percent fee — costs her about two thirds of her estate, more than ten percent extra spending does. And benefit timing, which generates the most debate anywhere retirement is discussed, moves her least of the five — about a tenth of her estate. For Priya specifically, that debate is not where the money is.

What to do with the ranking

Spend your attention at the top and stop worrying about the bottom. If benefit timing moves your answer by a rounding error, choose on non-financial grounds — when you want the income, how you feel about the risk — and move on. That is not laziness; it is allocating a limited amount of decision-making to where it changes something.

It also tells you which assumptions deserve scrutiny. An input the plan is highly sensitive to is one worth getting right; an input it barely notices can be left at a reasonable default without guilt.

The tweak tray moves each major input with a slider and the projection re-runs live, so the whole ranking exercise takes about as long as reading this episode did.

Run your own sensitivity pass

Next: where you retire changes the math — thirteen provinces, thirteen answers.

GlidePathEngine is an educational planning tool — not financial, investment, tax, or legal advice.