Stress testing your own plan

Most financial plans are built around a single expected outcome, which is the one thing that will certainly not happen.

Why testing matters

The problem with averages.

A projection using an average return produces a single number that conceals an enormous range of possible outcomes.

The actual sequence of returns will not be the average, and the order in which they arrive matters if you are withdrawing.

Which means a plan should be assessed on how it behaves across a range of scenarios rather than on whether it works in the central case.

The useful output is a sense of robustness rather than a precise figure.

The scenarios to test

Which cover the realistic range.

A large market decline early, particularly relevant near retirement, where a substantial fall in the first years of drawdown does disproportionate damage.

Lower returns than assumed, sustained over a long period rather than a single bad year.

Higher inflation than assumed, which erodes fixed income and increases required spending.

Loss of income, through redundancy, illness or reduced hours.

Living longer than average, which is the scenario that produces shortfall.

A major unexpected expense: care costs, a property problem, supporting a family member.

And a combination, since problems arrive together.

How to test them

Practically.

Take your current plan and re-run it with a lower assumed return — two percentage points lower is a reasonable stress.

Re-run with higher inflation.

Model a thirty per cent fall at a specific point and see what it does to the outcome.

Model losing one income for a year.

And model the plan running five years longer than expected.

Free calculators and spreadsheets do this adequately, and stochastic tools that model many sequences give a probability of success rather than a single answer, which is more honest.

What to look for in the results

The questions.

Does the plan fail entirely under any single stress, or does it degrade gracefully?

Which variable does it depend on most, which indicates where the fragility is?

Is there flexibility — spending that could be reduced, work that could continue, a date that could move?

How much margin is there before a change is required?

And what would the response actually be if the stress occurred?

A plan that requires everything to go approximately as expected is not robust.

Building in robustness

What helps.

Conservative assumptions, since a plan built on optimistic returns has a high probability of disappointment.

Flexibility in spending, which research consistently finds is the most effective single adjustment for retirement plans.

Flexibility in timing, since working a little longer has a disproportionate effect.

Guaranteed income covering essential spending, which means market outcomes affect discretionary spending only.

A cash buffer, which prevents forced selling.

Diversification, which reduces dependence on any single outcome.

And margin, which is the least exciting and most valuable input.

The household version

Not only investment.

Could the household manage on one income, and for how long?

What sick pay applies, and for how long?

What insurance is in place, and what does it actually pay?

What would be cut, in what order, and by how much?

How long would the emergency fund last at the essential expenditure level?

And what would happen if the main earner could not work again?

These questions are more consequential for most households than any portfolio question and are asked considerably less often.

Writing down the response

The output that matters.

For each stress, what will you actually do?

Written in advance, when calm, which converts a crisis decision into the execution of an existing plan.

Including specifics: which spending is cut first, which accounts are drawn on, who is contacted.

And including what you will not do, which is generally more important — specifically, that a market decline is not a reason to sell.

Reviewing the test

Periodically.

Annually, or after any significant change in circumstances.

Update the inputs with actual balances and current circumstances.

Re-run the stresses.

Check whether the margin has grown or shrunk.

And adjust contributions, timing or expectations accordingly, which is considerably easier done early than late.

The value of the exercise

Which is not the number.

The output of a stress test is not a precise probability, since the inputs are uncertain and the models are simplifications.

The value is in identifying which variables matter, where the fragility is, and what the response would be.

Which is knowledge that changes decisions, and which is available for the cost of an afternoon.

General information only, not investment advice. Projections are not guarantees and investments can fall in value. Consult a regulated financial adviser.