If One of Us Stopped Working: The Arithmetic
The question "should one of us stop working" used to arrive with a newborn, get settled against the daycare bill, and go away again when the youngest started school. That is still the version most financial writing answers. It is not the version I keep hearing.
The households asking me now have children of ten and thirteen. The daycare bill is long gone. What they have instead is a specific worry about a specific school, a job market that has stopped feeling like a safety net, and a sense that one of them is working to break even against costs that only exist because they are working. That is a different question with a different shape, and it deserves arithmetic rather than reassurance.
Three reasons this is getting asked now
School safety is a sustained worry, not a news cycle. Gallup's August 2025 survey found 41% of K-12 parents fear for their oldest child's physical safety at school — the fourth consecutive elevated year, against a long-run average of 34% since Gallup began asking in 1998, and not far off the 44% recorded three months after Uvalde. I am not going to editorialise about that figure. I will say that a worry four years deep stops being a reaction and starts being a planning input, and that families acting on it deserve to know what the action costs before they take it rather than after.
Homeschooling settled at a higher level. The National Home Education Research Institute puts roughly 3.4 million K-12 students in home education in 2024-25, about 6% of school-age children. Johns Hopkins' Homeschool Hub, which tracks the states publishing counts, has growth slowing hard — across the 22 states reporting for 2025-26, an increase of about 1.5%, down from about 6.3% the year before. Read those together and the story is not a boom. It is a step change that stuck, and a step change that stuck means a few million households made a decision with a salary attached to it.
The job is not always the thing being chosen. Some of the people running this scenario are not deciding whether to leave. They are finding out, in advance, what happens if they have to. That is the same calculation with the sign flipped, and it is worth doing while you still have the salary rather than in the fortnight after you lose it.
All three land in the same place: what does one income actually cost us. So let us do it properly, on a plan you can open and check.
The household
planbend ships with three finished sample plans. The mid-career one is the Hendricks family: Marcus and Priya, 44 and 42, North Carolina, two children aged 13 and 10. Marcus earns $7,400 a month in operations; Priya earns $5,750 in hospital administration. They have $473,400 across six accounts, a $232,000 mortgage, and two 529s.
Three hundred and forty-three dollars of slack. Hold on to that, because it is the number the whole decision runs into — and note how little it is for a household earning $163,800. Most of what looks like room in a budget is already committed to something.
Step one: stop the salary and change nothing else
Priya stops in January. Nothing else in the plan is touched.
Take-home falls $3,964 on a $5,750 gross — the gap between those two is the tax and the 403(b) deferral that stop along with the paycheck. Spending does not move at all, because nothing has told it to.
That is deliberate, and it is the design decision I care most about in this whole feature. An app that quietly zeroed the daycare line when a parent stopped working would produce a confident number that nobody could see or correct. Whether the second car goes away is a fact about your household, not a rule. So the app does not guess. It asks.
Step two: what was the job paying for?
The questions it asks are read out of the plan rather than printed from a template, and in this household two of the expected ones are simply absent:
There is no childcare to stop. Nora is 13 and Theo is 10. The daycare years ended years ago, and the single largest line in the conventional version of this calculation is worth zero. This is exactly why the teenage version of the question is harder than the toddler version — the cost that used to justify the decision is not there any more.
The health coverage does not move. Priya carries no payroll premium; Marcus's plan already covers the family at $310 a month. For a household where the stopping spouse holds the insurance this is usually the biggest unpleasant surprise, because the family tier on the other person's plan costs more than the employee tier they are on now. Here it is a non-event, and the app says so instead of inventing a number.
The 403(b) stops on its own. It is a percentage of a salary that no longer exists, so it goes to zero without anybody choosing it — and the match goes with it. That is reported rather than offered as a choice, which I think is the right way round: it is not a decision, it is a consequence.
What is left is the commute and everything attached to it. The app finds the lines a job pays for and asks what each becomes. Their answers:
Four hundred and thirty-seven dollars. People expect this number to be bigger. The gap is now −$3,184 a month, and the household has to find it somewhere.
Step three: the part I did not expect
There are two places to find it. You can cut spending or you can draw on savings, and I built the cutting one first on purpose, because a household staring at a gap does not reach for the brokerage before it reaches for the subscriptions. Offering the brokerage first quietly turns "can we afford this" into "how long can we fund it" without anybody choosing that.
So the app lists everything they could stop buying, biggest first. Eighteen lines, from Travel & Vacation at $407 down to coffee at $20.
Cancel the travel. Cancel the kids' sports. Cancel the charitable giving, the gifts, the pet budget, the gym, every subscription — all eighteen lines, to zero — and you are still $1,269 a month short. Not "tighten the belt", not "find some savings": every discretionary dollar this household spends, gone, and it does not get there. I did not design that outcome; it fell out of the plan, and it is the single most useful thing this feature has told me.
Something is deliberately not on that list, and it is worth saying why. This household puts $1,517 a month into savings and investments outside the 401(k). planbend counts that as committed money, not discretionary, so it does not appear among the things you could cancel — because "cut your spending" should never quietly mean "stop saving for retirement". Stopping it is a real option, and a big one. It is just a different decision, and the app makes you take it on purpose rather than by accident while scanning a list.
If they draw on cash instead of cutting, the app does that arithmetic too: $88,000 of liquid savings against a $3,184 monthly gap is 27 months of runway. Under three years, and then the question comes back without the savings.
Step four: what it costs at the end
Here is a thing worth understanding about every retirement projection, including mine. The engine only draws on accounts once you are retired. Before that, a household can run any deficit at all and the long-horizon number barely notices — it keeps funding the Roth, the brokerage and the 529s out of a paycheck that stopped.
Rather than teach the engine to guess where a shortfall comes from, the app asks that too: which account covers it. Pick the joint brokerage, from next January until retirement, and the plan finally answers the question it was asked.
Two hundred and twenty thousand off the retirement number. $749,439 off the end of the plan.
And then, writing this, I checked whether that number meant what it looked like. It does not, and the way it fails is worth more than the number was.
Over those seventeen years the plan is asked for $649,536. The joint brokerage holds $61,000. It comes up short in 2029 — three years in — and it is short in every year after that, with the annual contribution arriving each January and being drained the same year. Fifteen of the seventeen years cannot be funded. $534,507 of what the plan is instructed to withdraw never comes out, because a balance is floored at zero and the instruction simply does nothing.
So $749,439 is not the cost of covering the gap. It is the cost of the stopped contributions plus every dollar the brokerage had, and the real answer to "where does the money come from" is there isn't anywhere. Which is the same answer the cut list gave a moment ago, from the other direction: this household can just about absorb one income stopping by giving up everything discretionary, and it cannot absorb it out of savings, because the savings are three years deep and the decision is seventeen years long.
That is a harder answer than a big number, and a more useful one. It was also a bug in my own app, which I found by writing this paragraph. The engine floored the balance and threw the remainder away, so a draw the account could not carry produced exactly the same ending number as one it could afford, and nothing on the screen told you which you were looking at. It does now: if a withdrawal you set runs out of account, planbend names the year, names the account, and tells you how much of what you asked for never happened. Writing the post found it; the fix shipped before the post did.
None of which makes the decision wrong. It makes it a decision about the budget rather than about the portfolio, which is a different conversation and a much more concrete one. Some households will look at that cut list and find it obvious. Some will look at it and stop. Both readings are defensible. Only one of them is available to a household that never ran the numbers.
The version most people should actually run
Before anybody stops entirely, run half.
Roughly 56% of the damage for roughly half the hours, which is to say there is no cliff hiding in the middle of this decision — and no bargain either. What the arithmetic does not capture is that part-time often keeps benefits eligibility, keeps a résumé continuous, and keeps the return to full-time a conversation rather than a job search. If the trigger is a school year rather than a decade, the part-time version is frequently the whole answer.
Three things the plan cannot see
The salary on the way back in. A projection that resumes a career at the salary it was paused at is being optimistic, and mine does not know any better. If you expect to return lower, model the return as its own income change at the figure you actually expect. It is two boxes and it materially moves the answer.
The costs of the new arrangement. Homeschooling in particular is not free: curriculum, co-op fees, testing, activities that replace the ones school was providing, and a household that now uses heat and electricity all day. There is no default budget line for any of that, so add one. The whole point of the exercise is that the number is yours.
The Social Security record. Benefits are computed on 35 years of earnings, and years at zero are counted as zeros. A few of them matter less than people fear and more than nothing. You can see the shape of it in the Social Security calculator, and adjust the benefit in your plan if a long gap is likely.
And one more, which is not a limitation so much as the point: the plan has no field for whether your child is all right. It costs the decision. It does not make it.
Run it on your own household
That opens the real app against the sample family, fully unlocked, saving nothing anywhere. Go to the Scenarios tab, find Scenario Toggles, and press If one of us stopped working…. Pick a salary, pick a date, answer the questions it asks about your own household, and watch the gap appear. Then try the half-time version, and the one where the salary comes back three years later.
When you want the answer for your own numbers, planbend is free to start and does not ask for a bank login. If one income is the plan, the thing worth costing next is what your own health coverage would cost if the remaining job ever went away too. And if you would like a professional to look at it with you, there are directories on the Connect with a professional page.