Your Job Is the Biggest Position in Your Portfolio. I Diversified Everything Except the Thing That Pays for It.

I spent an evening earlier this year arguing with myself about whether my world ETF had got too American.

A whole evening, on a question about the smallest thing I own, inside the one fund I buy on the same day every month, automatically, whether or not I have an opinion that week.

Then I did the other sum, the one I had been skipping for about four years. My income diversification, if anyone wants to call it that, is a single line on a single bank statement. Every euro that buys those shares arrives from one employer, in one country, in one industry, paid in the same currency as the mortgage on the house we live in and the mortgage on the second one we are still paying off.

I had spread the small position across an entire developed-market index and left the big one at 100% of one thing.

That is the whole article. Your job is the largest, least diversified and most correlated position on your balance sheet, and no amount of fiddling with regional weights inside a fund gets near it. I became a boring investor the embarrassing way, by riding one lucky pick into a crypto cycle that sat me down and explained I was not a genius. That conversion fixed the part of my balance sheet I could see on a screen. It never touched the part that pays for it.

How big that position actually is

This is not a metaphor I invented on a train. It is how the accounting gets done at national level.

The World Bank publishes balance sheets for 151 countries and counts human capital, defined as the present value of the future earnings of the working population, as a formal asset class alongside roads, factories and forests. In their 2024 report it came out at 60% of total global wealth in 2020, and around 70% in high-income countries. Six times out of ten, when you add up everything a country owns, what you are adding up is people who still have to go to work on Monday.

The household version is blunter. In the Federal Reserve's 2022 survey, the median US family under 35 had a net worth of $39.0k against an income of $60.5k. Less than one year of pay saved. By the mid-forties the median family is still under three years of income, and even at 65 to 74, with the earning all but finished, it holds about 6.7 years of it. Put that next to the thirty-odd years of paychecks a thirty-year-old still has ahead, and the pile of money is not the main asset for a very long time.

Data: Federal Reserve, Survey of Consumer Finances 2022.

On my side of the Atlantic the anchor is Eurostat's: the expected duration of working life in the EU was 37.5 years in 2025. Multiply what you earn now by something in that neighbourhood and the answer stays bigger than your brokerage account for longer than feels comfortable.

So the forum argument about whether 12% or 15% emerging markets is right is tuning maybe a tenth of someone's total wealth while the other nine tenths sits in one place with one name on it.

Is your salary a stock or a bond?

Moshe Milevsky wrote a book on this and the test fits in one line: does your income move with the economy?

A tenured professor or a civil servant holds something that behaves like a bond. Reliable coupon, low volatility, barely any relationship to what the index did last quarter. A commission salesperson or a small-business owner holds something that behaves like equity with leverage attached.

Most of us are in between and guess wrong about which end.

Three questions, about a minute:

1. Did your pay, hours or bonus get cut in either of the last two downturns?
2. Is any part of what you earn tied to your employer's revenue, profit or share price?
3. Would a 20% drawdown in the index plausibly arrive in the same quarter as a hiring freeze at your company?

Three noes and you genuinely own a bond, and the honest conclusion is that you can probably carry more equity risk than you do. One or more yeses and your careful 60/40 is closer to 80/20 once you count the thing funding it.

I answered yes to the third without having to think about it. I work in software. I have watched the tone of budget conversations change within weeks of the market changing, and never once in the other direction.

The correlation nobody prices

Here is where it stops being an accounting curiosity.

The risk is not that any one of these fails. It is that they fail in the same quarter. Benzoni, Collin-Dufresne and Goldstein put maths behind it in 2007: wage growth and stock returns look barely related month to month and increasingly cointegrated over a decade. Run the correlation on a short window and it will lie to you, and you will feel diversified for about ten years before finding out you weren't.

The European version is a real town. Oulu, in northern Finland, was built around Nokia's handset business. When that went, the dismissals cascaded through the subcontractors, the region had more than 2,000 unemployed high-tech workers by the end of 2014, and by 2016 the city's unemployment rate was above 16%, roughly double the Finnish national average. Salary, sector, local employer base and local house prices moved at once, because they were never four things.

Ireland absorbed 40% of all EU big-tech redundancies in the big-tech layoff wave, for the boring reason that every American tech firm parked its EMEA headquarters in the same country. A reader in Dublin holding a world ETF was diversified across twenty-three developed markets and undiversified across exactly one. Germany is running the slow-motion edition, its car industry 42,300 people smaller in the first half of 2026 than a year earlier.

And that is before the house. Roughly half of all euro-area household wealth sits in the main residence, so your largest asset is a building standing in the labour market that pays your salary. When local prices fall, people stop being able to sell and move. The asset and the income hold hands on the way down.

People spend whole evenings arguing about whether their fund is too American, and I have argued that one myself, while almost everyone having the argument sits 100% home-weighted in the position that funds the fund.

Then 2026, which I want to handle carefully rather than dramatically. Announced US job cuts hit 1,206,374 in 2025, up 58%, and through August 2026 they are actually down 41% year on year. But technology alone is 29% of everything announced, and AI was cited in 116,175 announcements in the first eight months of 2026, about 22% of the total and more than the entire prior history of that category combined. And the caveat I would want if I were reading this: plenty of serious economists think the young-worker weakness behind those headlines is about interest rates and a post-pandemic correction, and Stanford itself found no aggregate displacement. Good. That strengthens the point. If the professionals cannot agree whether the next shock is AI, rates or tariffs, forecasting the cause is a losing game and fixing the concentration is the only move left on the board.

You cannot trade your job, which is the best argument against everything I just wrote

I have to deal with this properly, because it is the objection that could sink the piece, and the Bogleheads forum has been making it for fifteen years.

You cannot sell your future salary. You cannot short your employer without getting fired or, in some places, arrested. There is no market price and no rebalancing trade. A salary carries no contractual guarantee, so calling it a bond is sloppy, and the income-minus-expenses profile across a real life is lumpy enough that it is not cleanly any asset class at all.

All of that is correct. If an asset cannot be traded, then "diversify it" is not an instruction. It is a figure of speech.

So here is the distinction everything below rests on. Diversification is genuinely unavailable. Risk reduction is not. You cannot change the weight of the position, but you can change how much damage it does when it moves: how big the cash buffer is, whether you stack employer equity on employer salary, which skills you build, and how many careers the household runs. Four executable decisions on an untradeable asset, all available this month.

Why "seven income streams" is the wrong income diversification

The internet's answer to this problem is to add streams. Three to five, say the listicles. Most of that advice is wrong, and it is wrong in a way you can measure rather than argue about.

Bankrate's 2025 survey: the median side hustle produces $200 a month, down from $250 the year before, and 28% of side hustlers earn between $1 and $50 a month, which is the largest single band. Two hundred a month is $2,400 a year, about 3.4% of the US median family income of $70.3k. A stream that size does not move a concentration ratio. It does not round it.

41% of side hustlers spend it on discretionary purchases and only 28% save any of it, so the thing bought as insurance is mostly funding dinners. Participation also fell from 36% of US adults to 27% in one year. A second income that quits when work gets busy was never insurance.

There is a subtler failure too: most side hustles are the same bet twice, because freelance consulting in the industry that already employs you pays nothing in the one scenario where you need it, which is the industry itself going quiet.

Where the counterargument wins, and I will say it plainly: if your side work is a genuine second profession, in a different industry, that you intend to be able to live on, that is real diversification and better than anything in my portfolio. For everyone else the hours are worth more pointed at the big income than the small one, which is the argument I made at length when I paid for a coach instead of starting a shop.

What actually reduces the risk, ranked

Four levers, in the order I would spend effort on them.

1. Protect the occupation, not the employer. The displacement research is narrower than "be more employable". Workers who keep their occupation while changing sector recover; those forced into a different occupation take a persistent earnings hit, and skill redundancy puts people on permanently lower paths while skill shortage mismatches heal fast. So the thing worth buying is capability another industry would recognise, plus the relationships that produce an introduction next month. The test takes ten seconds: can you name three people outside your company who would take your call about a role this week? If not, that is the gap, and closing it is cheaper than recovering from it.

2. Size the emergency fund by your industry, not by a rule of thumb. "Three to six months" is one number applied to a distribution with a very long tail, and the tail varies by a factor of four. Among US workers who lost a long-held job between 2023 and 2025, the share still unemployed a year or more later was 8.1% in information, 23.1% in financial activities, 31.0% in professional and technical services and 36.5% in educational services. The classic safe white-collar résumé is in the slowest bucket.

Data: US Bureau of Labor Statistics, Worker Displacement 2023-2025, Table 4.

Do not import the American arithmetic if you are European, because unemployment insurance here is a different animal. The principle travels; the months do not. Ours is twelve months of essential spending, which is deliberately too big, and I now know it was sized by feeling rather than by my industry's actual re-hiring speed. The two landed in a similar place, which was luck.

3. Budget for the pay cut, not just the gap. This is the number that rearranged my thinking. Of people who lost a long-held full-time job in 2023 to 2025 and found another full-time job by January 2026, only 49% were earning as much as before, down from 62% two years earlier. The layoff is the event everybody pictures. The repricing is what actually moves the retirement date, and nobody models it. Run your projection at 85 to 90% of current income resuming after your industry-adjusted gap and look at what it does to the year.

Data: US Bureau of Labor Statistics, Worker Displacement 2023-2025.

Only 45% of those workers got any written advance notice, and the notice made no measurable difference to whether they found work again. The warning does not change the outcome. The buffer does.

4. Never stack employer equity on employer salary. I have never held RSUs, so I am borrowing this one. Inside retirement plans it mostly fixed itself: company-stock concentration above 20% fell from 14% of Vanguard participants in 2005 to 2% in 2023. It relocated rather than disappeared, and advisers now report senior tech employees holding 50 to 80% of household net worth in one employer ticker after a few years of vesting, against a common guideline of under 10 to 15% for any single position. Somebody holding 15% of their portfolio in employer stock has not taken a 15% position. They have taken a 15% explicit one on top of an implicit one spanning their whole career.

The European translation, for those of us with no equity comp, is closer to home: a house in the town where your employer is one of the largest employers is the same trade in bricks.

None of it works without the boring lever underneath, which is a savings rate high enough that leaving becomes a decision instead of an emergency. Bodie, Merton and Samuelson showed in 1992 that what lets young people carry equity risk is not youth but flexibility, the ability to retrain, move or work longer if things go badly. Cash is what buys that, and the savings rate is the lever you actually control.

The only income diversification that really works

Two careers in one household, in two different employers, ideally two different industries.

That is the entire list of things that genuinely uncorrelate a family's income, and it is not exotic: both spouses are employed in 49.1% of US married-couple families, rising to 66.3% of those with children, and full-time dual earning covers more than 70% of couples with children in Lithuania, Portugal, Slovenia and Sweden.

Which is awkward, because the standard FIRE script quietly dismantles it. The moment the numbers allow, one partner stops, and everyone treats that as the reward. Almost nobody writes it down as what it is, which is spending a diversification asset to buy time.

I am not telling anyone what to do inside their own house. Stopping may be exactly the right purchase. We have a small daughter and there is no version of this where I pretend two careers plus a newborn is free. It costs something real every week, and not all of it is money.

But it should be a purchase, made on purpose, priced out loud at the monthly money date rather than assumed away. The sentence I would want on the table is: if we stop this one, our household income becomes a single position, and here is what we are doing instead to make the remaining one durable. If there is no answer to the second half, the decision is not finished.

What I changed, and what I deliberately did not

No side hustle. No selling of the world ETF. Not one euro moved off the standing order, and nothing clever bought to hedge anything.

What changed is smaller and less satisfying. The emergency fund now has a reason attached to its size instead of a feeling. My learning hours go into things another industry would recognise rather than deeper into one company's internals, which is a different choice and slightly less fun. Our projection runs at 85% income on the way back rather than 100%. And I stopped pretending the regional-weights argument was the important one.

The test that started all of this takes five minutes and needs no tool. Write four things on one page: your employer, your industry, the currency your mortgage is in, and the country weight of your portfolio. Add your partner's employer if there is one. If three of those five say the same word, you are not holding four positions. You are holding one, wearing four labels, and the only reason it looks like four is that they live on four different screens. That split is most of why I ended up building a tracker that puts the salary, the houses and the portfolio on the same page, which is my own product, so weigh it accordingly. A sheet of paper does the same job.

Oulu recovered, for what it is worth. Tech employment there eventually passed its Nokia peak, rebuilt by ex-Nokia engineers who started companies. The people with portable skills and a cash buffer crossed the gap. The people without them did not.

That is income diversification for a salaried person. Not more streams. One income made harder to break, held by a household that can survive it breaking anyway.