A visual investigation · 1960–2026
The House, the Children, the Career — and Everything Life Forgot to Put in the Budget
The 30-Year Mortgage in the Age of the 3-Year Career
There is a house I want you to picture. Four bedrooms, two baths, a two-car garage, a lawn that needs more water than it deserves. It sits on a flat street in a Santa Clara County suburb, somewhere in the belt that runs from Sunnyvale through Santa Clara down into San Jose. It was built in the late 1950s by a developer who put up four hundred like it in eighteen months. It has been re-roofed twice, re-piped once, and painted a colour its original owners would not recognise.
It is not a special house. That is the point. It is the house at the centre of the ordinary American arrangement: two working parents, two children, a mortgage, a driveway, a decent school two blocks away. Between 1960 and 2026 it changed hands perhaps six times. Each time, a family stood in the empty living room and did the same arithmetic in their heads.
This essay is about that arithmetic. Specifically it asks a question that is harder than the one usually asked:
Could an ordinary professional family realistically build, maintain, and protect a middle-class life in the Bay Area — in 1960, in 1980, in 2000, and in 2026?
Notice the third verb. Protect. Almost every affordability analysis stops at “build”: can you close the purchase, can you make the payment. That is the easy half. The hard half is what happens over the following three decades, when the roof fails and the parent gets sick and the industry restructures and the market falls — usually not one at a time.
I want to be clear about what this is not. It is not an argument that one generation had it easy. Every generation in this story faced something genuinely brutal. The 1980 buyer paid a mortgage rate that would be considered a policy failure today. The 1990 buyer watched the defence industry hollow out around them. The 2000 buyer bought into a bubble that broke within eighteen months. The 2010 buyer had cheap money and no job security. Nobody in this story got a free ride.
The question is not who suffered more. It is a structural one:
How did the shape of the financial risk change over sixty-six years?
Not the size of it. The shape.
How to read the numbers in this essay
Long-run local housing data is genuinely patchy. Santa Clara County did not have a clean, continuous median-sale-price series in 1960, and pretending otherwise is how bad essays get written. So every figure here carries a tag, and the tag means exactly what it says.
A published figure from a named primary source — the Census Bureau, Freddie Mac, the BLS, the IRS, KFF, a county assessor. Linked in the Sources section. Reproducible.
Not directly published for this place and year, but reconstructed from adjacent published data — state or national figures, a neighbouring year, a documented index. The reconstruction method is stated.
A choice I made to keep the arithmetic consistent: 20% down, a 30-year fixed loan, maintenance at a stated rate. Change the assumption and the number changes. They are all stated.
An invented but plausible household used to make the arithmetic concrete. Not a real family, not a prediction, not a forecast. A worked example you can re-run with your own inputs.
Two rules I have tried to hold to.
One: every major historical dollar figure appears twice — in the money of its own day, and in 2026 dollars. A $19,000 house in 1960 is not a cheap house until you know what $19,000 was. It was about 2.6 years of a Santa Clara County household’s entire income, and roughly $214,000 in today’s money. Neither of those facts is trivial and neither should be hidden.
Two: where reliable Bay Area–specific data does not exist, I use California or national data and say so in the same breath. Precision that has been manufactured is worse than an honest range, because it is harder to argue with.
Part I
Eight buyers. One house. The same twenty per cent down and the same thirty-year fixed loan every time — so that what changes in the table is the world, not the financing strategy.
Here is the anchor. A representative four-bedroom single-family house in a Santa Clara County suburb: about 1,200 square feet in 1960, drifting up to about 2,100 by the 2000s as the housing stock modernised and buyers stopped tolerating one bathroom. Priced at roughly 1.1× the county median single-family sale price, because a four-bedroom in a good school attendance area has always traded above the median.
Every buyer in the table below puts twenty per cent down and takes a thirty-year fixed mortgage. Nobody uses an adjustable rate, nobody buys points, nobody does anything clever. That is deliberate. Holding the financing constant is what lets you see what actually moved.
The two ends of this series are solid. In May 2026 the California Association of Realtors put the Santa Clara County median single-family price at about $2.1 million; San Jose became the first US city to cross a $2 million median for single-family homes in 2024. Going backwards, a contemporaneous San Jose figure puts the median at $37,049 in 1975. The US Census decennial medians for owner-occupied homes nationally — $11,900 in 1960, $17,000 in 1970, $119,600 in 2000 — give the national spine.
The middle of the series is reconstructed from those anchors plus the Santa Clara premium over the California median, which widened substantially between 1980 and 2000. Estimate The 1960 and 1970 rows are the softest; treat them as accurate to about ±15%, which does not change a single conclusion in this essay.
← scroll the chart sideways →
| Decade | Purchase price | In 2026 $ | Median household income | In 2026 $ | Price ÷ income | Rate | Down payment | Principal | Monthly P&I |
|---|---|---|---|---|---|---|---|---|---|
| 1960 | $19,000 | $213,750 | $7,400 | $83,250 | 2.6 | 6.00% | $3,800 | $15,200 | $91 |
| 1970 | $28,000 | $240,309 | $12,700 | $108,997 | 2.2 | 8.50% | $5,600 | $22,400 | $172 |
| 1980 | $118,000 | $476,869 | $24,700 | $99,819 | 4.8 | 13.74% | $23,600 | $94,400 | $1,099 |
| 1990 | $285,000 | $726,129 | $48,100 | $122,550 | 5.9 | 10.13% | $57,000 | $228,000 | $2,023 |
| 2000 | $575,000 | $1,111,934 | $74,300 | $143,681 | 7.7 | 8.05% | $115,000 | $460,000 | $3,391 |
| 2010 | $625,000 | $954,264 | $89,100 | $136,040 | 7.0 | 4.69% | $125,000 | $500,000 | $2,590 |
| 2020 | $1,425,000 | $1,833,559 | $126,000 | $162,125 | 11.3 | 3.11% | $285,000 | $1,140,000 | $4,874 |
| 2026 | $2,300,000 | $2,300,000 | $175,000 | $175,000 | 13.1 | 6.71% | $460,000 | $1,840,000 | $11,885 |
← scroll the table sideways →
| Decade | Annual P&I | Property tax | Insurance | Maintenance | Total annual housing | % of gross income | P&I alone, % of income | Value of this house in 2026 |
|---|---|---|---|---|---|---|---|---|
| 1960 | $1,094 | $475 | $60 | $460 | $2,089 | 28% | 15% | $2,300,000 |
| 1970 | $2,067 | $700 | $110 | $730 | $3,607 | 28% | 16% | $2,300,000 |
| 1980 | $13,189 | $1,475 | $300 | $1,760 | $16,724 | 68% | 53% | $2,300,000 |
| 1990 | $24,274 | $3,562 | $600 | $3,120 | $31,556 | 66% | 50% | $2,300,000 |
| 2000 | $40,696 | $7,188 | $850 | $4,450 | $53,184 | 72% | 55% | $2,300,000 |
| 2010 | $31,083 | $7,812 | $1,100 | $5,910 | $45,905 | 52% | 35% | $2,300,000 |
| 2020 | $58,491 | $17,812 | $1,500 | $7,020 | $84,823 | 67% | 46% | $2,300,000 |
| 2026 | $142,624 | $28,750 | $2,200 | $9,030 | $182,604 | 104% | 81% | $2,300,000 |
← scroll the table sideways →
Assumptions behind Table 2. Modelling Property tax: before Proposition 13, an effective rate of 2.5% of market value, which is roughly where California sat in the mid-1970s. After 1978, 1.25% of the purchase price in year one — the Prop 13 base rate of 1% plus about 0.25% of voter-approved local bonds. Santa Clara County has over 900 tax-rate areas and real bills range from about 1.1% to 1.6%. Modelling Maintenance is deliberately not modelled as one per cent of the purchase price. Roofs do not get more expensive because land does. It is modelled as a physical cost — about $4.30 per square foot per year in 2026 dollars, deflated to each year by CPI and scaled to the size of the house. That is why the 2026 maintenance line is $9,030 and not $23,000. If you use the one-per-cent-of-price rule in the Bay Area you will over-reserve badly, because you are reserving against dirt. Estimate Insurance is a reconstructed series; the 2026 figure sits inside the $1,200–$2,400 range typical of low-wildfire-risk South Bay ZIP codes today.
← scroll the chart sideways →
The ratio did. In 1960 and 1970 the house cost between two and three years of the county’s median household income. By 2000 it cost nearly eight. By 2026 it costs thirteen. That is the single most important number in this essay and it is not subtle: the price-to-income ratio has multiplied roughly six-fold since 1970.
The down payment did. In 1970 the deposit on this house was about five months of median household income — the sort of thing a disciplined couple could assemble in three or four years. In 2026 it is two years and eight months of the entire gross income of a median household. Nobody assembles that out of salary. Which is a hint about where Bay Area down payments actually come from, and we will come back to it in Part VI.
← scroll the chart sideways →
The rate did not move in the direction people assume. Look at the two worst years for total housing cost as a share of income: 2000 at 72% and 2026 at 104%. Neither is a high-rate year by historical standards. Meanwhile 1980 — the year of the legendary 13.74% mortgage — comes in at 68%, which is better than 2000 and dramatically better than 2026. Hold that thought; it is Part III.
And one thing genuinely did not move: the house. Same footprint, same street, same school. Every row of Table 2 ends in the same number, $2,300,000, because it is the same house. What changed was not the asset. What changed was the claim that the asset makes on a life.
Before anyone gets nostalgic about 1960: that buyer paid an effective property tax rate of about 2.5% of market value, reassessed annually and upward, in an era with no cap and no protection. They had one income in most households, no index funds, no 401(k) — that did not exist until 1978 — a defined-benefit pension that depended entirely on the survival of a single employer, and a medical system that could bankrupt them without warning or recourse. Their house had one bathroom and no insulation worth the name. Their appliances broke constantly. Their real income was, in 2026 dollars, about $83,000 for a household. They were not comfortable. They were house-rich in a way that only looks obvious in hindsight, and cash-poor in a way that was obvious at the time.
Part II
The single largest reason two identical houses on the same street can carry property tax bills that differ by a factor of four.
In June 1978, California voters passed Proposition 13. In plain English it does four things. Fact
One. It caps the basic property tax rate at one per cent of assessed value. Local voter-approved bonds sit on top, which is why real bills in Santa Clara County run closer to 1.1–1.6%.
Two. It replaces market value with acquisition value. Your house is assessed at what you paid for it, not at what it is worth today.
Three. It caps the annual increase in that assessed value at two per cent, regardless of what the market does. In a decade when Bay Area prices rose eight per cent a year, an assessment rose two.
Four. The clock resets on change of ownership or new construction. When the house sells, it is reassessed at the new purchase price and a new thirty-year divergence begins.
The mechanism is simple. The consequence is not.
Bought 1995 · 31 years in the house
Effective rate on market value0.31%
Bought this year · next door
Effective rate on market value1.25%
The gap is $21,591 a year — roughly $1,800 a month, forever, for identical houses receiving identical services from the same county. Over a decade it is more than $215,000. That is not a rounding error in a household budget; it is a second car, a year of childcare for two, or four years of in-state tuition.
Now, the part that gets left out of most retellings.
Proposition 13 did something real and defensible. In the mid-1970s California homeowners — including retirees on fixed incomes — were being reassessed annually into tax bills that rose faster than their income, in some cases far faster. People were losing houses they had already paid for, to a tax on a gain they had not realised and could not spend. Prop 13 made the property tax predictable. If you own a house in California, you know today what your tax will be in 2040, within two per cent a year. Almost no other cost in your life has that property.
It also did something with consequences its authors did not fully price. Because assessed value only resets on sale, it creates a strong financial reason not to move — economists call this the lock-in effect, and it has been measured. It shifts the funding burden of schools, roads and fire services progressively onto whoever arrived most recently. And because commercial property changes hands less often than residential property, it shifted the relative burden between the two over time.
I am not going to argue for or against it here, and I would be suspicious of anyone who could settle it in a paragraph. What matters for this essay is narrower and undeniable:
Your property tax is not a market rate. It is a fixed toll on your entry date, and it is one of the largest single line items in your housing cost — $28,750 a year — close to the whole annual housing cost of the 1990 buyer ($31,556), and eight times what that buyer paid in property tax. When you compare your bill to your neighbour’s and it makes no sense, this is why. And when you model your own thirty-year cost, remember that this line grows at 2% a year, capped, which over thirty years is the friendliest line in your budget. Everything else compounds faster.
Part III
“We paid fourteen per cent in the eighties. Young people today have it easy.”
This sentence gets said at a lot of dinner tables, and it deserves better than the eye-roll it usually gets. Because the first half of it is true, and it was worse than people who did not live through it understand.
The Freddie Mac survey put the thirty-year fixed at an annual average of 13.74% in 1980 and 16.63% in 1981 — the highest annual average in the series. Fact Those were not headline rates for bad credit. That was the going rate for a good borrower with a job and a deposit. People bought houses at those rates because the alternative was not buying, and they did it while inflation was gnawing at their savings and the local economy was about to be reorganised.
So let us take the claim seriously and actually test it. Not with rhetoric — with the two things that determine a mortgage payment: the rate and the principal. And then a third thing that determines whether the payment is survivable: the income underneath it.
$118,000 house · 13.74% · 30-year fixed
Payment as share of median income53%
$2,300,000 house · 6.71% · 30-year fixed
Payment as share of median income81%
Look at the interest ratios first, because they are the most counter-intuitive line in this essay. The 1980 borrower pays 319% of their principal in interest — more than three dollars of interest for every dollar borrowed. The 2026 borrower pays 133%. On that measure, the eighties really were savage, and the complaint is entirely justified.
But percentages do not buy groceries. The 1980 borrower pays $301,284 of interest across thirty years. The 2026 borrower pays $2,438,716. Indexed at 1980 prices, the 1980 borrower’s total interest comes to roughly $1.2 million in today’s money — genuinely enormous, and still only half what the 2026 borrower pays. And that figure flatters 1980: those payments were made across three decades of high inflation, so each later dollar was worth less than the one before it. The real burden was lower again.
Then look at the ratio that actually decides whether a family survives: payment against income. 53% in 1980. 81% in 2026. On the only measure that a household treasurer cares about, 2026 is materially worse than the worst rate environment in the history of the survey.
Swap the rates. Put 1980’s 13.74% on today’s $1,840,000 loan and the payment becomes $21,424 a month — $257,000 a year, before tax, insurance or a single roof tile. Put today’s 6.71% on 1980’s $94,400 loan and the payment becomes $610 a month. Rates matter enormously. But they are multiplying very different numbers.
Find the rate that would make today feel like 1980. This is the question worth asking. What interest rate on a $1,840,000 loan would produce a payment equal to 53% of a 2026 median income — the same burden the 1980 buyer carried? The answer is 3.03%.
Which is, almost exactly, the rate of 2020.
That is the whole story of the last six years in one number. The 2020–21 rate window did not make Bay Area housing cheap. It made a $1.8 million loan behave like a 1980 mortgage. Take the window away, and the principal is the binding constraint. It always was; the rate was just hiding it.
Find the house instead. Hold the rate at today’s 6.71% and ask how big the loan would have to be to reproduce 1980’s burden. Answer: $1,205,000 — about a $1.5 million house. Which does exist in Santa Clara County. It is just not a four-bedroom on a good street.
The 1980 buyer’s 13.74% was a ceiling. Rates fell to 10.13% by 1990 and 8.05% by 2000, and every one of those steps was refinanceable. Many 1980 buyers refinanced two or three times and ended the decade paying substantially less than they started. Their worst month was their first month.
The 2026 buyer’s $1,840,000 is a floor. If rates fall to 4%, their payment drops to about $8,785 — real relief, and worth waiting for. But the principal does not move. You can refinance a rate. You cannot refinance a principal.
Interest rates matter. Principal matters too. Income relative to both matters most.
Part IV
How a household can be, simultaneously and truthfully, in the top decile of American earners and unable to buy an ordinary four-bedroom house.
Two hundred thousand dollars. Nationally, that is roughly the top ten per cent of household incomes. It is a number that, said out loud in most of the country, ends the conversation about whether you are doing well.
Here is that number meeting Santa Clara County. Scenario Married, filing jointly, two children under seventeen, both parents on W-2 income, standard deduction, no equity compensation, no inherited wealth, no 401(k) deferral yet because we are going to see whether there is room for one.
Housing alone consumes 94% of gross income and 124% of take-home pay. The household is $36,878 short before it has bought a single bag of groceries, insured a single car, or paid a single day of childcare.
A lender applying a conventional 43% back-end debt-to-income limit would allow about $7,167 a month of housing debt on this income. Principal, interest, tax and insurance here run $14,464 a month — 202% of that ceiling. No underwriter would write this loan. The paradox is not that the family cannot afford the house. It is that they are not even close, on an income that would be transformative almost anywhere else in the country.
This is the half of the analysis that alarmist versions of this article skip, and skipping it is dishonest. $200,000 in Santa Clara County is not poverty and pretending otherwise insults people who are actually struggling. Here is the real answer.
It can rent comfortably and save. A three-bedroom at $4,200 a month costs $50,400 a year, leaving $101,000 of take-home. That funds a lean but genuinely decent family life — the kind with swimming lessons and a summer camp and a trip to see grandparents — with modest saving on top. Not nothing. Not a house.
It can buy a house — a different house. Holding housing to 30% of gross, $200,000 supports a purchase of about $776,000; at an aggressive 35%, about $910,000. In Santa Clara County in 2026 that is a condominium, an older townhouse, or a small single-family home further out. Modelling It is a real home. It is not the four-bedroom on the good street, and the honest thing is to say that plainly rather than pretend the gap is a matter of budgeting harder.
What it cannot do is both. It cannot buy the representative house and fund the life that the house is supposed to be for. That is the paradox: the income is genuinely high, and it is genuinely insufficient for a specific and very widely held definition of arrival.
Which raises the obvious question. Somebody is buying these houses. They sell, month after month, at those prices. Who is buying them?
We will get there in Part VI. First, we need to see the rest of the budget — because the mortgage was never the expensive part.
Part V
The mortgage calculator assumes nobody gets sick, nobody gets laid off, the roof has signed a thirty-year maintenance agreement, and both children apparently plan to attend college for free.
Now the serious part.
Below is an annual budget for a family of four in Santa Clara County in 2026, in three columns. Lean is a household that is careful, drives older cars, cooks at home and says no to most things. Typical is the household most professional families actually recognise. Comfortable is what people mean when they describe a Bay Area professional life without irony.
Two rules for reading it. First, mortgage principal, interest and property tax are excluded, because those track whichever house you bought — they get added back in the scenarios that follow. Second, the children’s activities are not filed under frivolity. Swimming, an instrument, a sport, a martial art: this is the standard package that professional families in this county consider ordinary, and pricing it honestly is the point of the exercise. You may choose fewer. You should know what you are choosing between.
← scroll the chart sideways →
Modelling assumption These are illustrative ranges, not survey medians. They are built from published anchors where those exist — the KFF employer survey puts the average worker contribution for family cover at $6,850 in 2025, which is why the lean healthcare column starts there — and from ordinary Bay Area market prices where they do not. No family spends exactly this. The purpose is not to tell you what you spend. It is to make visible how many lines exist between a mortgage payment and a life, and how quickly the total gets past a quarter of a million dollars without anyone behaving extravagantly.
Childcare, before school starts. The table above assumes school-age children. It gets considerably worse first. Infant and toddler centre care in Santa Clara County runs about $27,400 per child per year — the highest in California. Fact Two children overlapping for a couple of years is $54,800 a year, more than the entire lean non-housing budget for food, transport and healthcare combined. Across the full pre-school phase for two staggered children, the total is somewhere north of $230,000. Estimate This is the expense that quietly decides whether the second parent keeps working, and it arrives at exactly the age when careers compound fastest.
College, eighteen years out. The published cost of attendance at a UC campus, living on campus, runs to roughly $42,000 a year in 2026; the average published cost of attendance at a private non-profit four-year college was in the low-to-mid $60,000s for 2025–26 (published tuition and fees alone averaged $45,000). Fact Escalate those at a conservative 4% a year and the arithmetic gets ugly fast.
| Destination | 4-year total, child starting in 10 years | 4-year total, child starting in 18 years | Two children | Monthly saving needed, 18 yrs at 6% |
|---|---|---|---|---|
| Public (UC) | $264,000 | $361,300 | $752,100 | $1,942 |
| Private | $411,500 | $563,200 | $1,172,400 | $3,027 |
Modelling Published sticker prices — $42,000 for a UC and $65,470 for a private non-profit — escalated at 4% annually, four years per child, second child starting two years after the first, saving compounded at 6%. Most families will not pay sticker — grants and institutional aid meaningfully reduce the net price for many, and the UC net price after aid is far lower. But a household earning $300,000 in Santa Clara County should plan on receiving very little need-based aid. Plan for the sticker; be pleasantly surprised.
Read those two numbers together. $1,942 a month for eighteen years, for public university, for two children. That is a second mortgage. It is also, almost exactly, the entire monthly principal-and-interest payment of the 1990 buyer — $2,023.
The reckoning
Six household incomes. Four housing choices. One question: after tax and shelter, is there enough left for the life the shelter was supposed to contain?
| Gross | Tax | Take-home | Rent a 3-bed · $4,200/mo | Buy $1.4M townhouse | Buy $1.8M 3-bed | Buy $2.3M 4-bed |
|---|---|---|---|---|---|---|
| $120,000 | $18,825 | $101,175 | $50,475 SHORT |
−$3,140 IMPOSSIBLE |
−$32,944 IMPOSSIBLE |
−$70,199 IMPOSSIBLE |
| $200,000 | $48,274 | $151,726 | $101,026 LEAN, NO SAVING |
$47,411 SHORT |
$17,607 SHORT |
−$19,648 IMPOSSIBLE |
| $300,000 | $83,908 | $216,092 | $165,392 LEAN + SAVING |
$111,777 LEAN + SAVING |
$81,973 LEAN, NO SAVING |
$44,718 SHORT |
| $400,000 | $120,858 | $279,142 | $228,442 LEAN + SAVING |
$174,827 LEAN + SAVING |
$145,023 LEAN + SAVING |
$107,768 LEAN, NO SAVING |
| $500,000 | $167,348 | $332,652 | $281,952 TYPICAL LIFE |
$228,337 LEAN + SAVING |
$198,533 LEAN + SAVING |
$161,278 LEAN + SAVING |
| $700,000 | $261,909 | $438,091 | $387,391 TYPICAL LIFE |
$333,777 TYPICAL LIFE |
$303,973 TYPICAL LIFE |
$266,717 TYPICAL LIFE |
← scroll the table sideways →
The residual covers the full $244,650 typical budget, including $80,000 a year of saving. The family lives the life the neighbourhood implies and funds its own future.
The residual covers the $111,240 lean budget, which includes $35,500 of saving. Careful, cooked-at-home, one activity per child. Retirement and college are being funded, slowly.
The residual covers lean spending ($75,740) but not lean saving. The family lives. The future does not get funded. This is the most common and most dangerous cell in the table.
The residual does not reach even lean spending, or housing exceeds take-home outright. Something in the picture has to give — the house, the location, or an assumption about what the family gets to have.
A few things fall out of this table that are worth sitting with.
The orange cells are the story. Not the red ones. Red is obvious — a family that cannot pay its mortgage finds out within months. Orange is the household that appears, from the street, to be doing everything right: they own the house, the children are in their activities, the cars are fine. They are simply not saving. Their retirement contribution is a token, their 529 is a gesture, their emergency fund is three weeks deep. They will discover this in year eleven, when something breaks.
Buying the four-bedroom requires roughly $500,000 of household income before the arithmetic stops being painful — and even at $500,000 it buys a lean life, not a comfortable one. That is the honest, unglamorous answer to the essay’s title question, at least for this specific house.
The tax wedge grows faster than you expect. At $120,000 the effective rate is 15.7%. At $700,000 it is 37.4%. Between those two points the household’s gross income multiplied by 5.8; its take-home multiplied by 4.3. That gap is not a complaint — progressive taxation is doing exactly what it is designed to do — but it is a planning fact. Doubling your salary does not double your capacity.
And renting is not the failure state. Look along the rent column. At $300,000, renting funds a lean life with saving. At $500,000 it funds the full typical life. In a market where the price-to-income ratio is thirteen, the household that rents and invests the difference is not losing a race. They are running a different one, and Part XI is about who actually finishes ahead.
Part VI
Salary rich and equity rich are not the same species, and the difference does not appear in a salary survey.
Return to the question we left hanging. If a $200,000 household cannot buy the representative house, and even $500,000 buys it only leanly, who is closing at $2.3 million?
Partly, households at $600,000 and above — there are a lot more of them in this county than in most. But mostly, the answer is not about income at all. It is about the down payment, and down payments in the Bay Area come from somewhere other than salary.
Consider two families. Scenario Same title, same company tier, same base salary of $250,000. Same age, same two children, same intention to buy the same house.
$250,000 base · no meaningful equity
Down payment capacity after six years$150,649
$250,000 base · $150,000 vesting annually
Down payment capacity after six years$793,940
Identical salaries. A 5.3× difference in purchasing power. And note what is not different: the recurring income that has to service the loan every month for thirty years is $250,000 for both families. The equity changed the deposit, not the payment.
This is why salary statistics are close to useless for explaining Bay Area housing. The visible variable — what people say when asked what they earn — is the one that moved least. The mechanisms that actually decide who buys are all invisible in that number:
None of these are shameful. They are simply the actual mechanism, and pretending the mechanism is salary makes people feel like failures for arithmetic reasons that have nothing to do with them.
Everything above makes equity compensation look like an unambiguous gift. It is not. It has a specific failure mode, and the failure mode is correlated — which is the word that matters.
Run Family B again with one change: the stock falls fifty per cent before the shares vest. Their incremental after-tax stream drops from $94,575 to $47,288 a year, and the six-year pile falls from $793,940 to $472,294 — a 41% haircut on the down payment. Scenario
That is survivable if it happens before you buy. It is a very different thing if it happens after, and if you used the RSU stream to justify the size of the mortgage. Because here is the asymmetry:
When the stock falls fifty per cent, the mortgage payment does not.
Worse, the failure modes arrive together. Stock prices fall when the company is struggling. Companies restructure when they are struggling. Layoffs happen during restructuring. Vesting schedules end and refresh grants shrink in exactly the quarters when the share price is depressed. The correlation between “my equity is worth less” and “my job is less secure” is not zero — it is close to one, and it points the wrong way.
Recurring salary should support recurring obligations. Variable compensation should primarily build resilience and wealth.
This is a risk-management posture, not a universal rule, and there are people for whom it is too conservative — a founder with genuine conviction, a household with a second income that could carry the whole mortgage alone, someone late enough in a career that the horizon is short. The point of the principle is not that it is always right. It is that the alternative — letting a variable stream service a fixed thirty-year obligation — is a bet you make once and cannot easily unwind.
| Approach | House | Down | Loan | Monthly P&I | All-in housing / yr | % of salary alone | Cash left over |
|---|---|---|---|---|---|---|---|
| Maximum house | $3,970,000 | $793,940 | $3,175,759 | $20,514 | $307,384 | 123% | $0 |
| Salary-supportable | $1,600,000 | $793,940 | $806,060 | $5,207 | $94,080 | 38% | $0 |
| House + reserve | $1,450,000 | $595,455 | $854,545 | $5,520 | $95,963 | 38% | $198,485 |
Modelling 6.71% thirty-year fixed; all-in housing is principal, interest, property tax at 1.25%, insurance and maintenance. The “% of salary alone” column deliberately ignores RSU income — that is the whole test.
The first row is what a lender will happily approve if the RSU history is strong enough, and it is how households end up with a mortgage that only works while the stock cooperates. The second row is defensible. The third row is the one I would want to be in: a slightly smaller house, and $198,485 sitting in reserve — which, as we are about to see, is the difference between a bad year and a catastrophe.
Part VII
Not as an apocalypse. As a duration problem.
I am going to be careful here, because this is the section where essays like this usually go off the rails in one of two directions: breathless claims that AI is about to eliminate everyone’s job, or dismissive claims that nothing is happening. Neither survives contact with the data, and the data is genuinely interesting.
Let me separate four things that get mashed together, because they have wildly different evidentiary weight.
The Stanford Digital Economy Lab, working with ADP payroll records covering roughly one-sixth of American workers — about 4.6 million workers across more than 730 occupations — has been tracking this since 2025. Their finding, as of mid-2026: there is no widespread, economy-wide job displacement associated with AI. Fact
But there is a concentrated one. Employment among workers aged 22–25 in highly AI-exposed occupations now sits about 19% below where it would be had it tracked similarly-aged workers in less-exposed occupations. That gap was 15% in July 2025 and 19% by June 2026 — it is widening, not stabilising. Fact
Two details matter enormously and are almost always dropped. First, the mechanism is reduced hiring, not increased firing. Companies are not dismissing juniors; they are not opening the requisition. Second, in occupations where AI is used to complement rather than automate, employment is flat or rising — particularly for experienced workers. This is not a uniform force. It has a direction and it has a target.
Challenger, Gray & Christmas track announced job cuts. Technology firms announced 149,023 cuts in 2026 through July, up 67% on the 89,251 announced over the same period in 2025. AI has been the single most-cited reason for five consecutive months, and appears in about 101,743 announcements year-to-date. Fact But read that carefully: these are announcements, and the stated reason is a corporate communication decision, not a measurement. “AI efficiency” is a more investor-friendly explanation for a cut than “we over-hired in 2021” or “demand fell”. Some of these cuts are genuinely AI-driven. Some are ordinary restructuring wearing a fashionable label. Nobody, including the companies, can currently tell you the split.
Surveys of worker anxiety measure sentiment, which is real and consequential — anxious workers negotiate differently, move less, and spend less — but sentiment is not employment. And economic projections of AI’s effect over the coming decade span a range so wide as to be nearly uninformative, because they depend on adoption assumptions nobody can currently ground. I am not going to present any forecast in this essay as a fact, and you should be sceptical of anyone who does.
So: no mass displacement. A real, measured, widening effect on entry-level hiring in exposed occupations. A corporate narrative running ahead of the evidence. And genuine uncertainty about the next ten years.
Here is why that adds up to a household finance problem anyway, and it has nothing to do with whether AI takes your job.
A thirty-year mortgage is a promise you make to a bank about the next three hundred and sixty months. It is one of the very few genuinely long-dated commitments an ordinary person ever signs, and it is completely inflexible: the payment is the payment, in the good years and the bad.
Against that, consider what a professional in 2026 can honestly say about five years out. Their employer? Uncertain — and in an industry that announced 149,023 job cuts in seven months, not idly uncertain. Their job title? Titles are being redefined faster than they are being filled. The market value of their current skill stack? Genuinely unknown; some skills that commanded a premium in 2021 command none now, and some that did not exist in 2021 command a great deal. Their compensation? Dependent on a share price and a refresh grant. Whether their function exists in its present form? For a meaningful set of roles, honestly unclear.
None of that requires AI to eliminate a single job. It only requires the cycle time of professional work to be shorter than the cycle time of the debt. And that has been true for a while — AI is accelerating something that outsourcing, offshoring, the 2001 bust, the 2008 crisis and the 2022–26 restructuring wave were each already doing.
The problem is not that the career is short. It is that the mortgage is long.
Every previous generation in Table 1 faced employment risk too — the 1990 buyer watched Bay Area defence and aerospace employment contract severely, and that was not gentle. What is different is the ratio. The 1990 buyer’s mortgage consumed 50% of median income and their house cost 5.9 years of it. The 2026 buyer’s consumes 81% and costs 13.1 years of it. Same uncertainty about employment; far less slack to absorb it.
That is the structural change this essay has been building toward. Not that things got more expensive — though they did. That long-dated fixed obligations grew, while the income streams underneath them became shorter-dated and more variable. Those two trends running in opposite directions is the actual risk. Everything in the rest of this essay is about what you do with that.
Part VIII
Budgets assume that life behaves itself. Life has never signed that agreement.
Everything so far has been a snapshot — a year, priced carefully. But a house is not a year. It is thirty of them, and thirty years is long enough for essentially everything to happen to essentially everyone.
What follows is not a prediction. It is a plausible sequence, and the specific items are less important than the pattern: the good thing and the bad thing arrive in the same five-year window, repeatedly, for four decades.
Buy the house
Then the org restructures four months after closing. One income stops. The mortgage does not. The house is nine weeks old and the emergency fund is being spent on it.
Children arrive
Then a parent becomes ill, eight thousand miles or eight hundred away. Flights, unpaid leave, a sibling who cannot help, and a decision about money that is really a decision about love. Childcare is running $54,800 a year at the same time.
Career progressing, promotion landed
Then the roof, the furnace and a slab leak arrive within one winter. $30,000, unbudgeted, non-negotiable, and no lender wants to hear about it.
College saving finally on track
Then a recession. The 529 is down 28% in the year you were counting on it, the taxable account is down more, and the natural instinct — stop contributing — is exactly the wrong one and completely understandable.
Peak earning years
Then career disruption — the function is reorganised, the skill stack is suddenly a decade old, and the job market for a $400,000 specialist is a great deal thinner than the job market for a $150,000 generalist. The search takes eleven months, not three.
Retirement accumulation, catch-up contributions
Then a child needs help — a deposit, a medical bill, a failed start, a divorce. You will help. You were always going to help. It was not in the model.
Mortgage nearly finished
Then a healthcare or eldercare expense that Medicare does not touch and long-term-care insurance was too expensive to buy in 2015. Memory care in this county runs into six figures a year.
Now — and this is the important turn — that list is not meant to be depressing. Read it again and notice what it does not say.
It does not say all of these will happen to you. It does not say any particular one will. It does not say the sequence is inevitable or that anyone did anything wrong. Most of these events are ordinary. Roofs fail. Parents age. Markets fall. Industries reorganise. If you live long enough and love enough people, some version of this list is simply what a life looks like.
The lesson is narrower, and it is the load-bearing sentence of this entire essay:
You cannot predict which crisis will arrive. You can build enough margin that one crisis does not destroy thirty years of work.
That distinction — between predicting the event and surviving the category of event — is the difference between a financial plan and a financial forecast. Forecasts are a form of entertainment. Margin is engineering.
Part IX
Five households. One bad eighteen months. Who is still standing, and what did it cost them?
Banks stress-test their balance sheets against adverse scenarios. Households almost never do, which is odd, because a household is a leveraged balance sheet with a single concentrated income source and no capital buffer requirement.
So let us run one. The scenario is deliberately harsh but not absurd — every element of it happened to somebody in 2001, in 2009, and in 2022. Scenario All of it happens at once, because that is how it actually goes:
Each household keeps paying only essentials: full housing cost, food, utilities, healthcare, transport, and the children’s basic needs. Every discretionary line is already cut to zero. No holidays, no camps, no restaurants, no saving of any kind. This is the floor, not a lifestyle.
| Household | Gross | All-in housing | Essential burn / yr | 18-month need | Income during shock | Cash gap | Reserves after the fall | Verdict |
|---|---|---|---|---|---|---|---|---|
| A · renting | $200,000 | $50,700 | $142,700 | $264,050 | $151,762 | $112,288 | $40,000 | ORANGE |
| B · $1.4M, 2022 | $300,000 | $96,740 | $194,740 | $342,109 | $208,222 | $133,887 | $77,500 | ORANGE |
| C · $1.8M, 2024 | $400,000 | $143,321 | $263,321 | $444,982 | $257,266 | $187,716 | $162,000 | ORANGE |
| D · $2.3M, 2026 | $500,000 | $182,974 | $322,974 | $534,461 | $309,951 | $224,509 | $325,000 | YELLOW |
| E · $3.0M, 2026 | $700,000 | $223,504 | $388,504 | $632,756 | $409,255 | $223,501 | $790,000 | GREEN |
← scroll the table sideways →
Modelling Households A–C have no RSUs or small ones; D has $150,000 vesting annually and E has $300,000, both halved in the shock. Reserves are cash plus taxable equities after the 30% fall. Retirement accounts are excluded from “reserves” because reaching them early costs tax plus a 10% penalty — a household that has to touch them has already lost. These verdicts are directional, not precise. Change the reserve assumption by $50,000 and colours change. That sensitivity is itself the finding.
The shock is absorbed from cash and taxable investments with reserve left over. Nothing structural changes. The family has a rough year and keeps its plan.
Survivable, but the emergency fund and the taxable account are emptied. Recovery takes three to five years of rebuilding. One more shock in that window is very bad.
Reserves do not cover the gap. Requires raiding retirement accounts, taking on debt, or a significant restructuring — sell a car, pull children from activities, stop the 529, refinance or forbear.
Forced sale of the house, or serious debt on unfavourable terms, at the worst possible moment in the market. This is the outcome the entire essay exists to help you avoid.
Three findings from this table, and none of them are the obvious one.
Income does not buy resilience. Margin does. Household C earns twice what A earns and lands in the same colour. Household E, on $700,000, is the only green — not because of the salary but because they held $300,000 in cash and $700,000 in taxable investments. Household D, on $500,000 in a more expensive house, is yellow. The variable that separates the colours is not income. It is the ratio of reserves to essential burn.
The house drives the burn, and the burn drives everything. Household E’s essential spending is $388,504 a year — nearly three times household A’s. That is a lot of machine to keep running. Halve E’s reserves and the green turns yellow. Cut them to a quarter — still $197,500, more than most households will ever hold in cash — and E lands in the same orange as a family earning half as much. The $700,000 income does not rescue them, because the shock removes a piece of that income by construction.
Everybody is thinner than they think. Look at how few months of essential spending each household’s cash covers.
← scroll the chart sideways →
| Household | Essential burn / mo | 6 months | 12 months | 18 months | 24 months | Cash held | Months covered |
|---|---|---|---|---|---|---|---|
| A · renting | $11,892 | $71,350 | $142,700 | $214,050 | $285,400 | $40,000 | 3.4 |
| B · $1.4M | $16,228 | $97,370 | $194,740 | $292,109 | $389,479 | $60,000 | 3.7 |
| C · $1.8M | $21,943 | $131,661 | $263,321 | $394,982 | $526,643 | $120,000 | 5.5 |
| D · $2.3M | $26,914 | $161,487 | $322,974 | $484,461 | $645,948 | $150,000 | 5.6 |
| E · $3.0M | $32,375 | $194,252 | $388,504 | $582,756 | $777,008 | $300,000 | 9.3 |
← scroll the table sideways →
Every one of these households would describe themselves as having an emergency fund. Four of the five have less than six months of essential spending in cash, and every one of them would tell you they are conservative with money. They are not lying. Their burn rate is simply larger than their intuition about it, because the burn rate is set by a house they bought once, years ago, in an afternoon.
This is the single most useful calculation in this essay and it takes four minutes: add up your genuinely non-optional annual spending, divide by twelve, and divide your accessible cash by that number. The answer is usually smaller than expected and it is the only number that matters when the bad eighteen months arrive.
Part X
You don’t predict the event. You build resilience.
An underwriter is answering a narrow question: will this loan probably perform? They are not asking whether your marriage survives it, whether your children get the activities you wanted for them, or whether you can take a lower-paying job in eight years without selling the house. Approval is a floor on plausibility, not a recommendation.
“Can we still carry this house if one income disappears for a year?”
If the answer is no, the house is too big — not morally, just arithmetically. And notice this is a much harder test than the payment test, because it is about the whole housing cost including tax, insurance and the roof reserve, against one income after tax.
This is Part VI restated as a rule. Salary is a recurring stream with a notice period. RSUs, bonuses, and equity appreciation are episodic streams whose value is set by a market and whose continuation is at someone else’s discretion. A mortgage is permanent. Match the durations.
Which does not mean equity compensation is wasted. It means directing it at things that increase optionality rather than things that consume it: the emergency fund, taxable investments, the 529, principal reduction, the retirement accounts, and the fund in principle four.
The conventional advice is three to six months of expenses. That advice was formed in a labour market where a competent professional could replace a job in weeks and where household burn rates were a fraction of what Table 7 shows. It is worth asking whether it still applies to a highly specialised Bay Area professional earning $400,000 in a narrow function.
I am not going to declare a single correct answer, because there isn’t one. But here is the shape of the trade-off:
| Reserve | What it covers | Who it suits | What it costs you |
|---|---|---|---|
| 6 months | A clean job change with a short gap. One moderate shock, alone. | Two-income households in broad, liquid job markets; renters; households with no dependants. | Very little. This is the minimum, not a target. |
| 12 months | A genuine search in a soft market, or one shock plus a repair. | Most two-income professional families with children and a mortgage. | Roughly one year of essential spending sitting in cash, earning less than equities over time. |
| 18 months | The full stress test in Part IX, for most of these households. | Single-income households; specialised senior roles; equity-heavy compensation; anyone supporting parents. | Real opportunity cost. Over a decade the foregone return is meaningful — and so is not being forced to sell equities at the bottom. |
| 24 months | A career transition, a retraining year, or a sabbatical taken deliberately. | Founders, people whose function is being restructured, households near a planned change. | Significant drag if held permanently. Best held for a defined period around a known risk. |
There is no universally correct row. The honest way to choose is to ask how long your replacement takes — not how long it took in 2021, when everyone was hiring, but how long it would take in the market you would actually face if your industry were the one that had contracted.
A single “savings” account does two things badly: it hides how many separate obligations it is meant to cover, and it makes every withdrawal feel like the same decision. Named buckets fix both. They are a psychological device with a real effect on behaviour.
6–18 months of essentials
Cash or near-cash. Boring by design. This is not an investment; it is insurance you self-underwrite.
$6,000–$10,000 a year
Roof, HVAC, plumbing, sewer, exterior. Funded monthly so the $30,000 winter is an annoyance rather than an event.
Family out-of-pocket max
An HSA where available. One serious illness should not compete with the mortgage for the same dollars.
$1,000–$3,000 a month
529s. Started early, because Table 3 is unforgiving and the compounding does most of the work in the first decade.
To the cap, first
$24,500 per earner in 2026, plus the match. The one bucket where the tax code is actively on your side.
$4,000–$7,000 a year
Because two cars in the Bay Area will both need replacing, and financing a depreciating asset at the wrong moment is how good budgets break.
Whatever is honest
The bucket most people fund emotionally and not financially. If you know you will help, fund it. If you know you cannot, say so early.
Small and protected
Not frivolous. A family that never rests makes worse decisions, and the resentment compounds faster than the savings.
12–24 months of essential burn · separate from the emergency fund
This is the one that almost nobody has and that changes the most. The emergency fund exists so that a bad thing does not destroy you. The opportunity fund exists so that a good thing does not pass you by — and so that a bad thing can be met with a choice instead of a surrender.
It is what lets you spend nine months learning a genuinely new skill instead of taking the first job offered. Take a sabbatical you will not get another chance at. Start the company at thirty-eight rather than never. Accept a role that pays 30% less because it is where the next decade is going. Relocate. Change careers entirely at forty-five. Sit out a bad market instead of selling into it.
Every one of those is a wealth-creating move that is unavailable to a household with no slack — and unavailability is the real cost of a maximum mortgage. It does not show up as a loss. It shows up as a choice you never got to make.
Employment is a state someone else controls. Employability is a stock you own, and it depreciates if unmaintained. In a labour market where entry-level hiring in exposed occupations is measurably down and the corporate story is about efficiency, the thing worth compounding is the ability to be hired — by someone, somewhere, at a wage that services the mortgage.
In practice: continuous learning that is actually current rather than nostalgic; genuine AI literacy, which means knowing what the tools do badly as well as what they do well; a professional network maintained before you need it; cross-functional range so that a single function’s decline is not your decline; financial flexibility so you can accept a lateral move; and, where possible, more than one way to earn.
Note that the last item on that list is financial. Career resilience and financial resilience are the same system. Money buys the time to retrain; retraining protects the money.
Return to household C from the stress test — $400,000 income, $1.8 million house bought in 2024, orange. Here is the same family with one thing changed.
| Version | All-in housing | Essential burn | Cash gap in the stress test | Reserves | Months of cash | Verdict |
|---|---|---|---|---|---|---|
| As bought: $1.8M, 5.5 months of reserve | $143,321 | $263,321 | $187,716 | $162,000 | 5.5 | ORANGE |
| Bought $1.45M instead, same deposit | $112,399 | $232,399 | $141,333 | $162,000 | 6.2 | YELLOW |
| Bought $1.8M, but held 12 months of reserve | $143,321 | $263,321 | $187,716 | $307,000 | 12.1 | YELLOW |
Two entirely different routes to the same improvement. The second row buys less house. The third row buys the same house and delays the purchase by roughly two years to build the reserve. Modelling Both move the family out of the zone where a bad eighteen months forces a sale. Neither requires earning more.
And notice the second-order effect in the middle row. The smaller house frees $30,922 a year, which is not just a cushion — it is the difference between a household saving 6% of take-home before discretionary spending and one saving 17%. Over twenty years, at that gap, the smaller house is very plausibly the wealthier outcome, before we even count the crises it survives.
Part XI
Two households. Almost identical net worth. One of them is seven and a half times freer than the other.
$600,000 compensation · $3.0M house
Years of financial freedom1.4 years
$300,000 income · $1.4M house
Years of financial freedom10.5 years
← scroll the chart sideways →
Net worth says these households are equivalent — within three per cent of each other, which is inside the noise of any real valuation. On the conventional scoreboard, Household A is arguably ahead: twice the income, a house worth more than twice as much, and a retirement balance more than double.
Now measure something else.
Years of Financial Freedom
How long can this household maintain its essential life with no employment income at all? Accessible assets divided by essential annual burn. Nothing more complicated than that.
Household A: 1.4 years. Sixteen months. If they cut every discretionary dollar to the bone, 1.6 years. If they raid the retirement accounts and eat the tax and the penalty, four years.
Household B: 10.5 years. Cut to the bone, 13.4 years. They do not need to touch retirement at all.
Same net worth. Seven and a half times the freedom.
The mechanism is not mysterious and it is not about virtue. Household A earns twice as much and has 2.4 times the essential burn, because a $3 million house at 6.5% with a $1.8 million mortgage costs $185,627 a year to inhabit, before food. Their wealth is real, but most of it is either inside the thing they live in or locked behind a penalty. The portion that answers the phone in a crisis is $500,000, against a burn of $364,427.
Household B made a smaller house work, refinanced into a 4% mortgage and did not trade up, and put the difference into liquid investments for fifteen years. Their house is less impressive. Their reserve does the job that Household A’s house cannot: it converts a career shock from a crisis into an inconvenience.
Household A is not doing anything wrong, and there are futures in which they win decisively — if their compensation holds for fifteen more years, if the $3 million house appreciates faster than B’s portfolio, if nothing goes wrong. That is a real scenario and it is not rare in this county. The point is not that A made a mistake. It is that A is running a strategy that requires the next fifteen years to cooperate, and B is running one that does not. Those are different risk positions wearing the same net worth.
Net worth tells you what you own. Years of financial freedom tells you what you can survive.
I would go further and say this is the number most professional families should be tracking instead of net worth, because it is the one that changes when you make the decisions that actually matter — how much house, how much cash, how much of the compensation is recurring. Net worth goes up when the market goes up. Freedom goes up when you choose it.
Part XII
Everything in one place. Not to award a medal for suffering — to show that the risk did not simply grow. It changed shape.
| Decade | House price | Median income | Price ÷ income | Rate | Down payment | Monthly P&I | Property tax | Non-housing family cost | College burden | Retirement burden | Career environment | Major economic risk | This house in 2026 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1960 | $19,000 | $7,400 | 2.6 | 6.00% | $3,800 | $91 | $475 | $6,700 | UC charged residents no tuition — fees only. Effectively negligible. | Employer pension plus Social Security. Household saves little by design. | One employer, one career. Orchards turning into electronics. | 1960–61 recession. All retirement risk concentrated in one firm’s survival. | $2,300,000 |
| 1970 | $28,000 | $12,700 | 2.2 | 8.50% | $5,600 | $172 | $700 | $8,800 | Tuition introduced for residents; still nominal in today’s terms. | Pension still the norm. The 401(k) does not yet exist. | Semiconductors arrive. Employment stable, mobility rising. | Stagflation. Two oil shocks. Savings eroded by inflation. | $2,300,000 |
| 1980 | $118,000 | $24,700 | 4.8 | 13.74% | $23,600 | $1,099 | $1,475 | $18,700 | Public tuition rising but a state school remains reachable on one income. | The 401(k) era begins. Responsibility starts shifting to the household. | PC boom. Job-hopping becomes normal and then expected. | Rates at 13.7%, then 16.6% in 1981. Back-to-back recessions. | $2,300,000 |
| 1990 | $285,000 | $48,100 | 5.9 | 10.13% | $57,000 | $2,023 | $3,562 | $29,700 | Public tuition climbing faster than wages. Saving early starts to matter. | Defined-benefit plans closing. 401(k) becomes the primary vehicle. | Defence and aerospace contract sharply. The Valley reinvents itself. | 1990–91 recession. Local prices flat to falling for years. | $2,300,000 |
| 2000 | $575,000 | $74,300 | 7.7 | 8.05% | $115,000 | $3,391 | $7,188 | $39,100 | 529 plans arrive. Private college costs decisively outpace inflation. | Fully on the household. Market risk is now personal risk. | Dot-com peak. Equity compensation becomes central to Bay Area pay. | The crash. Options worthless, jobs gone, the house bought at the top. | $2,300,000 |
| 2010 | $625,000 | $89,100 | 7.0 | 4.69% | $125,000 | $2,590 | $7,812 | $49,500 | Public tuition roughly triples over the decade. Student debt becomes structural. | Household-borne, plus the memory of a 50% drawdown two years earlier. | Recovery. Mobile and cloud hiring boom. Cheap money everywhere. | The aftermath. Local prices had fallen steeply from the 2007 peak. | $2,300,000 |
| 2020 | $1,425,000 | $126,000 | 11.3 | 3.11% | $285,000 | $4,874 | $17,812 | $58,700 | Full cost of attendance at a UC pushes past $35,000 a year. | Household-borne. Contribution caps rising slower than the cost of living here. | Pandemic, remote work, a hiring frenzy, then the 2022 correction. | The cheapest money in history, capitalised straight into the price. | $2,300,000 |
| 2026 | $2,300,000 | $175,000 | 13.1 | 6.71% | $460,000 | $11,885 | $28,750 | $75,700 | UC tuition and fees $15,588; full cost near $42,000. Private near $65,000. | Entirely household-borne. $24,500 per earner, against a $2.3M house. | AI restructuring. Entry-level hiring in exposed roles measurably down. | Rate normalisation meeting a price set at 3%. Duration mismatch. | $2,300,000 |
← scroll the table sideways →
Modelling “Non-housing family cost” is the lean 2026 budget of $75,740 deflated to each year by CPI — a deliberately crude device, and it understates how much cheaper childcare and healthcare were in real terms before 1990, and how much more expensive they are now. Read it as an order of magnitude, not a measurement. Estimate The qualitative columns are characterisations of well-documented conditions, not quantitative claims.
It does not say earlier generations had it easy. Read the two rightmost qualitative columns down the page. Every single row contains a genuine catastrophe. The 1960 household’s entire retirement depended on one company not failing — a risk that has since been distributed to the individual, which is worse in some ways and much better in others. The 1980 buyer signed at 13.74% and then watched rates go to 16.63% and two recessions arrive back to back. The 1990 buyer bought into a regional economy that was about to lose much of its defence and aerospace employment, and then watched their house go sideways for half a decade. The 2000 buyer bought at the exact top of a bubble and lost their options, their job and their paper wealth inside eighteen months.
And it does not say this generation has it easy either. Today’s buyer has tools nobody before them had: index funds at three basis points, tax-advantaged accounts that did not exist in 1970, a global labour market, portable healthcare that is not chained to one employer, and information that used to be locked inside brokerages. Those are enormous advantages and it is dishonest to wave them away.
What the table says is narrower and more useful than either grievance:
In 1960, the biggest financial risk to a family was a single catastrophic event — the plant closes, the pension fails, someone gets sick without insurance. The house itself was a small, stable claim on income: 2.6 years of it, 28% a year to run.
In 2026, the biggest financial risk is a duration mismatch. The house is a 13.1-year claim on income, financed over thirty years, serviced by employment whose visible horizon is measured in single digits, and partly funded by compensation whose value is set daily by a market. Nothing has to fail catastrophically. The pieces simply have to fall out of alignment for eighteen months.
That is a different kind of problem, and it does not respond to the same solutions. You cannot insure against a duration mismatch. You can only carry enough slack to survive one.
In conclusion
So — can you really afford the Bay Area?
I am not going to tell you not to buy a house. Owning the roof over your children is one of the more sensible things a person can do with money, and the households in this essay who own are not fools. Nor am I going to tell you that AI will take your job; the evidence does not support that and the people claiming it are mostly selling something.
And I am certainly not going to tell you that previous generations had it easy. Read Table 10 again. Nobody in it had it easy.
What I want to leave you with is a different definition of the word in the title.
Affordability is not: can I make next month’s payment? That question is answered by an underwriter in four minutes and it is nearly useless, because it assumes a world in which every month resembles this one.
Affordability is: can my family continue to live with dignity when life does not follow the plan? When the smaller income stops for eighteen months. When the roof and the parent and the market arrive in the same year. When the function you have spent fifteen years mastering is reorganised out of the org chart by people who have never met you.
A house is supposed to provide security. It should not require thirty years of uninterrupted good luck to keep.
And if that is right, then wealth is measured wrong almost everywhere. Not by salary, which is only the visible fraction. Not by house value, which you cannot spend without becoming homeless. Not by RSUs, which are a claim on someone else’s future. Not by cars, which are a claim on nothing. Not even, quite, by net worth — because Household A and Household B had the same net worth and only one of them could survive a bad decade.
Measure it instead by the thing this essay kept circling back to:
How much freedom do you have when the paycheck unexpectedly stops?
The real American dream may not be owning the biggest house.
It may be owning your time.
Method and sources
Three things are worth saying plainly before the list.
First, the county price series before 1980 is reconstructed, not published. Santa Clara County does not have a clean continuous median-sale-price series reaching back to 1960. The 1960 and 1970 rows are built from national and state decennial census medians plus the documented Santa Clara premium, and they are the softest numbers in the essay. If they are wrong by fifteen per cent in either direction, every conclusion here survives unchanged — the ratios move by a rounding error.
Second, the tax model is a model. It uses 2026 federal brackets, the 2026 standard deduction, the 2026 child tax credit, the 2026 Social Security wage base, California’s 2026 SDI rate, and California’s progressive brackets. It ignores itemised deductions, AMT, capital gains, state credits beyond the exemption credits, and the substantially different treatment of equity compensation. It is right to within a few thousand dollars for a straightforward W-2 household, which is all it needs to be. It is not tax advice and should not be used as such.
Third, none of this is financial advice. It is arithmetic, done in public, with the assumptions written down so you can disagree with them specifically rather than generally. Every household is different, and the ones in this essay are invented.
If you want to re-run this for your own household, the four numbers that do the most work are these. One: your all-in annual housing cost — principal, interest, property tax, insurance, and a real maintenance reserve, not just the mortgage. Two: your genuinely non-optional annual spending, everything else stripped out. Three: your accessible assets — cash plus taxable investments, excluding retirement accounts and home equity. Four: three divided by two.
That last number is your years of financial freedom. It is worth knowing, and it takes about four minutes to work out. Most people are surprised, and the surprise is almost always in the same direction.