Second-earner insurance

Why do households keep a second earner whose pay barely covers the childcare it requires?

A household sits down with a spreadsheet during the nursery years and finds something deflating: after childcare, commuting, tax and the extra convenience food, the second job clears almost nothing. Sometimes it clears less than nothing. And yet the household usually keeps it.

The tempting reading is that they have not done the sums, or that they have and are being stubborn. But this pattern is common enough, across enough income levels, that we should suspect the sums are being asked the wrong question rather than answered wrongly. What question would make keeping the job the sensible answer?

Begin with the arithmetic itself, because part of the puzzle dissolves there. Notice how childcare was charged: entirely against the second wage. Why? The children require care because both adults work, so the cost belongs to the household, not to one earner. Split it evenly and neither job looks like a loss-maker. That is not a trick — it is a correction. The habit of subtracting the whole of childcare from whichever wage is smaller makes the smaller wage look optional by construction, which is a conclusion smuggled into the accounting rather than derived from it.

But suppose we do the accounting fairly and the second job still clears very little. Is it then irrational to keep it? Only if the value of a job is its net cash this year. Ask instead: what is the household buying that it would not otherwise have?

Consider what happens if the main earner loses their job. A single-earner household's income goes to zero and stays there until they find work — weeks, often months, during which fixed costs do not pause. A two-earner household's income falls by the larger share but does not vanish; there is still a wage arriving, and the mortgage keeps being paid while the search runs. The second job's cash contribution in a normal year was near zero. Its contribution in a bad year is the difference between a squeeze and a crisis.

That is a redundancy argument, and it is worth stating in its own terms. A redundant component is one that adds almost nothing to output while everything is working, and adds everything when something fails. Judging it by its contribution on a normal day guarantees you will conclude it is waste — which is precisely the mistake the spreadsheet was making. The second income is not a marginal earner. It is a second circuit.

Now push on the obvious objection: why not leave the labour market and return if a shock arrives? Two reasons, and the first is well documented. Skills, contacts and currency decay during an absence, and the earnings penalty for a career break persists long after the return — the literature on the "child penalty," notably Henrik Kleven and colleagues' work across several countries, finds long-lasting earnings gaps opening at the birth of the first child. So a job re-entered after five years is not the same job in wage terms. The second reason is timing: shocks do not wait. Job-hunting takes months, and the household needs income during those months, not after them.

There is a supporting piece of evidence here that is easy to miss. Economists have long looked for the "added-worker effect" — a spouse entering work when the main earner loses theirs — and generally find it modest. That weakness is not a counterexample to the insurance story; it is a confirmation of it. If entering the labour market at the moment of need worked well, holding a job in advance would be unnecessary. The fact that late entry is a poor substitute is exactly why staying attached is worth paying for.

One more failure mode belongs in the list, though it is uncomfortable to say aloud: separation. A member with no recent employment history has weaker options if the partnership ends — and keeping a job preserves an exit that has nothing to do with intending to use it.

This does not settle every case. When childcare genuinely exceeds the household's discretionary income, or the work is unstable or damaging, stepping out can be the better call. The claim is only that a calculation counting this year's net cash cannot see what the second job is for.

The most useful correction here is about what a spreadsheet is for. Net cash this year is an expected-value calculation, and expected value is the wrong tool for a decision whose payoff lives in the tail. A household is not trying to maximise average income; it is trying to avoid ruin while doing reasonably well. Once that is the objective, an option that trims average income while cutting the variance can be a good buy at a price that looks absurd on the average.

It is also worth separating insurance from ambition. The insurance case for the second job holds even if the job leads nowhere, but most second jobs also carry a second, larger return: the wage after the childcare years end. The expensive period is short — a handful of years per child — and the career it protects runs for decades. Charging a temporary cost against a permanent asset makes those few years look decisive when they are, in the long run, a bridge.

And note that none of this requires the second earner to be the lower-paid one, or a particular gender. The argument is structural. It applies to whichever job the household is tempted to treat as the marginal one — which is usually the one the childcare bill happened to be subtracted from.

• How subsidised or capped childcare changes the calculation, and why countries with it show smaller and shorter earnings penalties after a birth.
• Whether the same redundancy logic justifies an emergency fund instead, and how the two compare pound for pound.
• The unequal distribution of the premium: who in the household actually pays it in hours and sleep.

Every term the collection defines is gathered in the glossary.