Rules of thumb, and where they break

Every rule of thumb about withdrawals is a model with the assumptions deleted.

That is not a criticism of the people who use them. A rule that fits in one sentence is enormously more useful than a spreadsheet nobody opens. It is a criticism of what happens next, which is that the sentence gets repeated for twenty years while the assumptions underneath it are never stated, checked or argued about.

This lesson does not give you a rate. It gives you the four things any rate is quietly claiming, and a way to check each of them.

Take four percent as the example, because it is round

Not because it is right, and not because this course is recommending it. It is a convenient illustration and every number below moves when the assumptions do.

Four percent of a pile, adjusted each year for inflation, over a thirty-year horizon. That is the shape. Now take the four assertions in turn.

The return. The rule needs the portfolio to earn enough, in real terms, after every cost, to fund the withdrawal for the whole period. Real, after fees, after tax where it applies. The gap between a headline return and that figure is routinely one to two points, and one to two points is the entire margin.

The horizon. Thirty years is a choice. Someone stopping work at fifty may need fifty. A rate that works over thirty and fails over fifty is not a safe rate that occasionally breaks; it is a rate for thirty years being used for something else.

Flat spending. Real spending is not flat. It tends to be higher early, lower in the middle, and higher again at the end when care costs arrive. A rule assuming a flat line is wrong in both directions and does not cancel out.

The order. The one people skip.

Predict

Two portfolios earn exactly the same average real return over thirty years. One has its worst years at the start, the other at the end. Both fund the same withdrawals. What happens?

Choose one answer

Nothing — the average is the same, so the outcome is the sameThe one with bad years first can run out while the other finishes with a large surplusThe one with bad years first ends slightly lowerIt depends on the size of the withdrawals

Lock in answer

Breaking the rule on purpose

The model starts with assumptions that make four percent look comfortable over thirty years. Two controls: the real return, and the fee that comes off it.

Withdrawal sustainability

≈ Hypothetical model, not a forecast

How long a portfolio lasts at a chosen withdrawal rate, and the rate that would last forever.

It assumes

• The same real return arrives every single year, in the same amount.
• The withdrawal is taken at the start of the year and holds its purchasing power.
• The fee is charged on the return every year, in good years and bad.
• No tax is paid, nothing is inherited, and no income arrives from outside.

It ignores

• The order in which returns arrive, which matters enormously here
• Tax on withdrawals, on income and on gains
• Spending that changes with age, health or circumstance
• Any income from outside the portfolio
• How long the person doing the withdrawing actually lives

Leaving the withdrawal at four percent, find assumptions under which this pile empties before year thirty. Then ask whether those assumptions are unreasonable.

Real return4%

Annual fee0.3%

Taken out each year

40,000

Real return after the fee

3.7%

Balance at the horizon

760,968

Total taken out

1,200,000

Year the portfolio empties

30

Portfolio empties inside the horizon

0

Rate the portfolio could sustain forever

3.568%

Rate that lasts exactly to the horizon

5.375%

• Balance
• Taken out to date

◷ Not there yet — keep moving the controls.

Somewhere around a net real return of one point three, thirty years stops working. That is a fall of a couple of points from the opening figure — well inside the range of what a long period can deliver — and it does not require any disaster.

The lesson is not that four percent is too high or too low. It is that the number is an output. Change two inputs nobody can observe and the same rate goes from comfortable to ruinous, which is a property of every rate, including whichever one you had in mind.

Two readings

Use a fixed rule

Set a rate at the start, adjust only for inflation, and do not react to markets.

Measured in

Whether the plan survived the historical periods it was tested against

For example

Drawing the same real amount every year for twenty years without recalculating.

Rests on

That people execute simple rules and abandon complex ones, so a slightly worse rule that gets followed beats a better one that does not.

Chase it alone and

It cannot respond to a bad first decade, which is precisely the scenario that ends plans. The rule is silent exactly when information arrives.

Use a flexible rule

Recalculate periodically, and reduce spending after poor years or when the balance falls below a threshold.

Measured in

Chance of running out, and how much spending had to fall in the bad cases

For example

Skipping the inflation increase in any year following a fall in the portfolio.

Rests on

That a household can genuinely cut spending when asked to, and will do so promptly rather than assuming the market will recover.

Chase it alone and

It moves the risk from running out of money to having much less to live on, and the size of the required cut is rarely stated in advance.

Both sides accept
Both sides agree that no fixed rate is safe in all conditions, that costs and horizon change the answer materially, and that the order returns arrive in matters more than the average.

Which is which? Put each one on a side.

• A household whose spending is almost entirely fixed commitments

  Use a fixed ruleUse a flexible rule

• A household with a large discretionary share and a long horizon

  Use a fixed ruleUse a flexible rule

• Someone who will not look at the portfolio more than once a year

  Use a fixed ruleUse a flexible rule

What to do with all of this

Not pick a better number. Numbers are the output.

The practice that survives contact with reality is to write the four assertions down as assertions — this return, this horizon, this spending path, this tolerance for a bad first decade — and revisit them on a schedule. When one turns out to be wrong, the rate changes, and it changes because something observable moved rather than because of a feeling about the news.

That is a duller answer than a rule. It is also the only one that stays true when the assumptions do not.

Check

Somebody says a particular withdrawal rate is safe. What is the most useful reply?

Choose one answer

Safe over what horizon, after what costs, and assuming what happens if the first five years are bad?No rate is safeThat rate was derived from one country over one periodIt depends on your risk tolerance

Check answer

Worth remembering

• Every withdrawal rule asserts a return, a horizon, a spending path, and that the order of returns does not matter.
• The order does matter: withdrawals during an early fall sell units that are never there for the recovery.
• A rate is an output of its assumptions, so assume a couple of points less real return and a comfortable rate becomes a ruinous one.
• Fees and tax come off the real return, which is the figure the whole rule depends on.
• Fixed rules survive because they get followed; flexible rules survive because they respond, and each fails where the other works.