The $800 Corporation

Eleven years, one Seychelles corporation, zero payroll tax: the legal architecture, the real costs, and what it does to your Social Security check.

The $800 Corporation

This one is for Americans: US citizens and green-card holders. If you're neither, your payroll taxes work differently, and this post will only entertain you.

In January my Seychelles corporation goes into voluntary liquidation. It cost $800 a year, never had an office, and for eleven years it did one job: it kept about $345,000 of self-employment tax in our brokerage account instead of sending it to the Treasury. Stacked on the foreign earned income exclusion, which did the larger share of the work, and invested as it arrived, the money they kept between them has compounded past a million dollars. Nothing was hidden. Every entity appeared on our tax returns, every form arrived on time. The engine was two taxes: one every expat knows, and one most nomads believe they're stuck with.

The familiar one is the foreign earned income exclusion (FEIE), which wipes out US income tax on your first $132,900 of foreign earnings in 2026. The unfamiliar part is the FEIE's hole, and for self-employed nomads the hole costs about $14,000 a year. The exclusion does nothing to self-employment tax. The exclusion lives in one chapter of the tax code; self-employment tax lives in another, and the exclusion never reaches it. A self-employed nomad netting $100,000 abroad owes zero income tax and $14,130 in self-employment tax: 15.3% of 92.35% of net earnings, every year, no matter where in the world the work happened. Section 1402 taxes US citizens on worldwide self-employment income whether you're in Raleigh or Kuala Lumpur.

Meanwhile, an American employed by a foreign company, working outside the United States, pays nothing into Social Security. Not the employee half, not the employer half. Nothing.

The entire strategy is the distance between those two sentences.

If you're a W-2 employee of an American company working from Da Nang, you're on the wrong side of that distance. FICA follows your paycheck anywhere on earth, and short of a certificate of coverage under a totalization agreement, nothing you file changes it. You're not stuck, though; you're one move away. Persuade your employer to convert you into a genuine contractor, with real deliverables and no daily supervision, because a relabeled employee is a misclassification suit their lawyers already know about. Or quit and sell the same skills to other clients. Either route lands you where this post picks up: self-employed, holding the $14,000 problem and everything that fixes it.

The rule that opens the door

Sort by who owes these taxes, because the sorting is the whole game. Work physically in the United States and FICA applies no matter who signs the paycheck. Work abroad as an employee of an American company and FICA still follows you. Work abroad for yourself and §1402 taxes your worldwide self-employment income, as the nomad with the $14,130 bill just learned. One person is left standing: the US citizen working outside the United States as an employee of a foreign employer.

Look at the three variables in that sentence. Where you work. Who employs you. Whether the employer can be a company you own. You control all three, and for a self-employed nomad, changing the middle one is mostly paperwork.

The statutory language, from §3121(b), defines FICA "employment" outside the US as only:

any service, of whatever nature, performed outside the United States by a citizen or resident of the United States as an employee for an American employer

"American employer" is a defined term in §3121(h): a US-organized corporation, a US-resident individual, a partnership where two-thirds or more of the partners are US residents, or the US government. A corporation organized in the Bahamas or Hong Kong is none of these, even if you own every share of it.

So a US citizen working abroad as an employee of a foreign corporation earns wages that are not FICA "employment." And wages aren't self-employment income under §1402, so self-employment tax doesn't apply either. Both doors close at once.

The corporation you form yourself counts. Nothing in the statute requires the foreign employer to be unrelated to you. That's not an oversight the IRS hasn't noticed; it's the published position. The IRS's own page on persons employed by a foreign employer states the exemption plainly.

Ownership doesn't change the answer either. Congress used residence tests for the other prongs of §3121(h): an individual employer who resides in the US, a partnership with two-thirds US-resident partners. For corporations it wrote place of incorporation, full stop, with no look-through for who owns the shares. And Congress knows how to write a look-through when it wants one: §3121(z) treats a foreign entity as an American employer only when it's part of a domestically controlled group performing US government contracts. For everyone else, a Seychelles corporation owned 100% by an American is still a Seychelles corporation. What your ownership creates is a controlled foreign corporation, and that's a reporting problem, not a payroll-tax problem. One check on the way in: the entity has to count as a corporation under the IRS's own classification rules, because a disregarded entity can't employ its owner.

The structure

The mechanics, in order:

You form a corporation in a foreign jurisdiction. Pick the jurisdiction on three criteria, in this order: zero or near-zero local corporate income tax, banking access, and annual maintenance cost. Hong Kong offers respectable banking and demands an audit every year; Singapore exempts small companies from audit but requires a resident director. A one-person services company needs neither. Mine is in the Seychelles, administered by Fidelity Corporate Services (no relation to the brokerage; they handle registered-agent work across many offshore jurisdictions) for about $800 a year, all-in. I have no relationship with them beyond eleven years as a customer.

Banking is easier than the offshore mystique suggests. Fintechs will sometimes let you add a business account for a foreign company to an existing personal profile; Wise did this for me years ago, and its list of supported jurisdictions changes, so check it before you incorporate anywhere. No suitcase of documents, no in-person account opening in a marble lobby. There is no prize for prestige here.

The corporation employs you under a written employment agreement, and you perform all services while physically outside the United States. If your clients can pay the foreign corporation directly, they do. If you have an existing US entity that clients pay, the US entity signs a management services contract with the foreign corporation and pays it a fee for your work.

That second configuration is mine. My US company, an S corporation, faces the customers: contracts, invoices, credit cards, the American banking relationships clients expect. The Seychelles company signed a contract with the US company to produce the work product, and my wife and I have worked for the Seychelles company from wherever we are. Customers never see the foreign entity. The work crosses the ocean; the money follows it under the services contract; the wages come out the other side.

Anyone who has run an S corporation will ask the obvious question: why not stop there? Pay yourself a modest salary, take the rest as distributions, skip the payroll tax on the distributions, no Seychelles required. Two reasons. The salary an S corporation pays you is a wage from an American employer, and §3121(b) reaches those wages anywhere on earth, so FICA applies to every dollar of it. And the distributions aren't earned income under §911, so the FEIE never touches them; they're taxed in full. The S corporation alone gets you payroll tax on the salary and income tax on everything else. Adding the foreign employer flips both: a full FEIE-cap salary with no payroll tax, and a US company that no longer pays anyone to work.

That last part protects the S corporation too. The IRS's standard attack on an S corporation is that the shareholders underpaid themselves in wages and took the difference as distributions. The defense here is that the shareholders perform no substantive services for the S corporation. It holds the contracts and the bank accounts and buys the work, under a services agreement, from a company that employs the people who do it. Keep it that way. The day you start doing the work for the S corporation directly, its distributions start looking like wages, and wages from an American employer are the one thing this structure exists to avoid.

That intercompany fee is where amateurs get hurt. Section 482 lets the IRS reallocate income between commonly controlled entities when pricing isn't arm's length. The fee the US entity pays the foreign corporation has to look like what it would pay an unrelated contractor for the same services: documented, commercially reasonable, with the foreign corporation keeping a modest profit, because real businesses have profits. For a contract service provider that owns no customer relationships and carries no business risk, the defensible price is its costs plus a modest markup, which is exactly what a token profit is. Put the pricing in writing when you set it up; contemporaneous documentation is the difference between an adjustment and an adjustment plus penalties. And the blade cuts both ways: the Service can just as easily argue your US company underpaid itself and move income back onshore.

The foreign corporation pays you wages. Those wages are earned income for FEIE purposes; §911 cares about where you performed the services, not who signed the paycheck. Keep the salary at or under the $132,900 exclusion and you've paid no income tax and no payroll tax on it.

One footnote: §3121(l) lets an American employer that owns at least 10% of a foreign affiliate file an agreement with the IRS, on Form 2032, bringing the affiliate's US-citizen employees into Social Security, and the agreement is irrevocable. It exists for multinationals that want to keep expat employees in the system. Nobody files one by accident, but know it exists.

What "outside the United States" actually means

Physical presence while performing the services. That's the whole test, and it's a bright line with no grace period.

The general rule of §3121(b) makes any service performed within the United States employment, regardless of who the employer is. Fly home for July, answer client email from your sister's kitchen in Charlotte, and those are US workdays: FICA applies to the wages allocable to them, your foreign corporation technically has US payroll obligations it is poorly equipped to meet, and the same wages are US-source income the FEIE can't exclude. Do enough of it and the problem graduates: the foreign corporation itself can become engaged in a US trade or business, with a US corporate return, corporate tax, and a 30% branch profits tax on top waiting at the end of that road. The corporation should stay out of the country even more scrupulously than you do.

The clean answer is the honest one: don't work in the United States. Vacation there. If you must work during a US visit, account for those days and expect them to be taxed like the domestic wages they are.

One more geography problem: totalization agreements. The US has roughly 30 bilateral Social Security agreements, and if you reside in an agreement country (most of Western Europe, Japan, South Korea, Australia among them), the agreement decides which country's system you contribute to. Escaping FICA into French social charges is not a victory. And the corporation has its own geography: manage it for long enough from one country and that country can decide the corporation is resident there, or has a permanent establishment there, and tax it as a local company with local payroll obligations attached. Perpetual movement keeps you, and it, out of host-country systems in practice; settling in Lisbon on a residence visa does not, and several of those systems take a larger bite than the 15.3% you left behind. The itinerant worker, the perpetual traveler, whatever you want to call the person who never stays long enough to become anyone's tax resident: that's who this structure was practically built for.

The FEIE has its own geography test, and it's about your abode, not your suitcase. An itinerant's tax home travels with the work, so a true nomad usually has a foreign tax home. But §911(d)(3) denies the exclusion to anyone whose abode stays in the United States, and abode means the center of your personal life: the house you kept, the family living in it, the place you fly back to every eight weeks. The half-out nomad loses the FEIE in Tax Court; the one who actually left doesn't. You also still have to pass one of the two qualifying tests: bona fide foreign residence, or 330 full days in foreign countries, not merely outside the United States, in any twelve months. Days at sea and days in Antarctica count for neither side, which matters if you cruise. The payroll-tax exemption doesn't care about any of this, so even the half-out nomad escapes FICA on foreign workdays. The income-tax half of the structure belongs only to people who actually left.

And the federal statute can't fix your state. Florida has no income tax, which is one reason we're domiciled there. California doesn't recognize the FEIE at all, and a Californian who never formally left gets the payroll-tax half of this and none of the income-tax half.

Work this doesn't fit

No profession is banned from the structure. The problem cases are about arrangement, not occupation.

The worst one hides in §954(c)(1)(H). If a client contract designates you personally as the person who must perform the services, or gives the client the right to say who will, and you own 25% or more of the corporation, the corporation's income from that contract becomes foreign personal holding company income: subpart F income, taxed to you currently at ordinary rates. The payroll-tax exemption survives, but any hope of leaving profit in the corporation dies. If your clients are hiring you by name rather than buying services from your company, structure the contracts so the company commits to deliver the work and the client has no say in who does it, and understand the line you're walking.

The second problem is substance, and the thing being tested is the employment itself. This whole structure rests on you actually being an employee of the corporation; if the "employer" exercises no control, keeps no payroll records, and exists only on a registered agent's shelf, the IRS argues you were self-employed all along and the corporation was your agent, and §1402 walks back in the door. The doctrine is older than the statute: income is taxed to the person who earns it, and a paper entity doesn't earn anything. The fix is boring diligence. A real employment agreement, real payroll with wage statements, real minutes, government fees paid on time, and the part paper can't supply: a corporation that actually holds the client contracts and directs the work.

And work that requires your body in the United States, in-person consulting, US court appearances, a speaking circuit, breaks the geography requirement for exactly those earnings. Hybrid years are possible but messy.

The GILTI problem, which now has a different name

Your foreign corporation is a controlled foreign corporation the day you form it, and since 2018 the profits you leave inside a CFC no longer wait politely offshore. GILTI taxed them to you currently. As of January 1, 2026, GILTI is gone in name: the One Big Beautiful Bill Act renamed it Net CFC Tested Income, cut the deduction that set its rate, and repealed the 10% deemed return on tangible assets that used to shelter the first slice. The base got broader and the rate went up. For an individual owner who makes no special elections, that income lands at your ordinary rates, and the FEIE can't touch it because it isn't earned income.

For a one-person services corporation the answer is almost embarrassingly simple: don't leave profit in the corporation. Salary is deductible against tested income. Pay yourself essentially everything as wages, keep the token profit that makes the company look like a business, and the NCTI inclusion rounds toward zero. In one recent year my foreign corporation grossed about $350,000, deducted salaries and expenses of $346,000, and closed the books on $3,347 of profit. That's the entire GILTI problem, solved on line 19 of an income statement. You still report everything, every year; you just owe almost nothing on it.

One technical note on the leftover sliver. Services a foreign corporation performs for a related person, outside its home country, are foreign base company services income under §954(e): subpart F, an older and harsher regime than NCTI, and a related-party services company like mine sits squarely inside it. Don't reach for the de minimis rule. It tests gross income, and when every dollar the company takes in is a related-party service fee, the company fails it completely and lands in full inclusion instead. What saves you is the same thing that saves you on NCTI: subpart F income is computed net of the deductions allocable to it and capped at earnings and profits, so the inclusion is the profit, not the receipts. My $350,000 year produced $3,347 of subpart F income, taxed at my ordinary rates, and left nothing over for NCTI to reach. Two regimes, one sliver, same answer. If real profit ever accumulates, a §962 election lets you compute the tax as if a US corporation held the shares. That's a conversation to have with someone who files these forms for a living.

The calibration that makes all of this run clean: size the services contract so the foreign corporation takes in roughly what it pays out in salaries plus that markup, and set the salaries at the FEIE limit. Nobody strains reasonable-compensation doctrine paying a working professional $132,900; the salary has to be defensible as pay for actual services, and at FEIE-cap levels it almost always is. Everything the business earns beyond that never needs to touch the foreign corporation at all. It stays in the S corporation and passes through to us on a Schedule K-1 as ordinary income: taxed at our rates, but never subject to self-employment tax, and, because we materially participate in the business, not to the 3.8% net investment income tax either.

Then the multiplier. The exclusion is per person, and so is the payroll-tax exemption, so a couple who both work in the business gets two of everything: two FEIE-cap salaries from the foreign corporation, $265,800 of wages in 2026, income-tax free and payroll-tax free, and a second $18,800 a year of self-employment tax that never comes due, about $37,600 for the household. That is where the structure stops being a nice optimization and starts funding a brokerage account. Nothing in the statute prices the second employee any differently from the first; the corporation just has two names on the payroll and a services contract sized for both.

The condition is the same as everything else here: the second spouse has to actually do the work. A spouse on the payroll who contributes nothing is the arrangement examiners are trained to find, and it fails three ways at once: the salary isn't reasonable compensation, the services fee that funds it isn't arm's length under §482, and the employment the exemption depends on doesn't exist. If your spouse doesn't work in the business, the ceiling is one exclusion and one exemption, which is still the $14,000 problem solved.

What the structure costs

Form 5471, every year, because you're a controlling US shareholder of a foreign corporation. The penalty for not filing starts at $10,000 per year and climbs. Worse than the penalty: under §6501(c)(8), the statute of limitations on your entire return doesn't start running until the 5471 is filed. Skip it and, absent reasonable cause, every year in between stays open to audit indefinitely. Over a ten-year horizon, budgeting for one penalty event somewhere is realism, not pessimism. Add the FBAR once your foreign accounts top $10,000 in aggregate, Form 8938 above its own thresholds, and an accountant who actually knows international filings.

Here are my real numbers, because the internet is full of estimates from people who haven't filed one. The Seychelles company costs about $800 a year to maintain. My personal return, Form 5471 and all its schedules included, costs $750 to prepare. The corporate return for my US company runs about the same, but I'd pay that with or without the foreign corporation. The incremental overhead is under $2,000 a year.

At 15.3%, $2,000 of overhead eats the savings on the first $13,000 of earnings. The structure starts making sense somewhere north of $40,000 in net earnings and gets better fast. On my own numbers, about $18,800 a year in self-employment tax on a 2026 FEIE-cap salary, and $16,000 a year averaged over the eleven years the cap was climbing, times two for my wife and me, the overhead is noise.

Which is the argument for keeping it cheap, and the offshore industry will fight you on it. The providers who advertise to nomads sell packages: a Dubai free-zone company with a residence visa, a Singapore entity with a nominee director and an annual audit, a Cayman structure with a management company attached, $8,000 to $15,000 a year before the accountant. None of it buys anything the statute cares about. Section 3121(h) asks where the corporation was organized; it does not award points for prestige, a local office, or a director you've never met. At $12,000 a year of overhead, the structure doesn't break even until $85,000 of earnings, and a two-earner household at the caps has handed a third of its savings to the people who set it up. A registered agent, a shelf-company jurisdiction nobody has heard of, and a fintech bank account is the whole build. Anything beyond that is the provider's margin, not your protection.

Plan the exit before the entrance. Move back to the United States and the structure dies on arrival: no foreign workdays, no FEIE, and a CFC you no longer need. Wind it down carefully, because liquidating a foreign corporation with accumulated earnings can convert them into a taxable dividend under §1248. The clean version is the one this whole post describes: a corporation that never accumulated anything worth taxing on the way out. Mine goes into voluntary liquidation in January, eleven years in, with nothing inside it but the share capital we put there.

The other end: what you'll collect

Every promotional foreign-corp article skips this part. Opting out means your Social Security earnings record goes quiet, and you should know exactly what that costs before you sign anything.

Your benefit is built from your highest 35 years of indexed earnings. Miss years and zeros fill the gaps. The average feeds a formula with three brackets, and the brackets are the entire story. For 2026 eligibility, the formula pays 90% of your first $1,286 of average indexed monthly earnings, 32% of the next slice up to $7,749, and 15% of everything above that.

Read that again from the bottom. If you already have 35 years on your record and your average indexed earnings run above about $93,000 a year, which is where the second bend point sits, your next contribution dollar buys benefits at the 15% rate. One more self-employed year at $132,900 credits your record with 92.35% of it, raises the 35-year average by about $292 a month, and lifts your check by about $44 a month. Obtaining that $44 costs $15,219 in the Social Security portion of SE tax. You break even 29 years into retirement. The Medicare portion, another $3,559, buys a marginal benefit of exactly nothing if you already have your 40 quarters.

Run a full career through the formula and watch what happens. Take someone who averaged $150,000 a year, in today's wage-indexed dollars, from 23 to 53: thirty solid years. At 53 they take the show on the road, form the foreign corporation, and contribute nothing for five years. Five zeros drop into their 35-year average.

Staying in for those five years would have credited five more years at $138,525 each, the 92.35% of net earnings that counts toward the record, for a benefit of about $3,918 a month. Opting out drops it to about $3,670. The zeros cost $247 a month, $2,967 a year, roughly $59,300 over a 20-year retirement.

What did they avoid paying? Self-employment tax on $150,000 runs $21,194 a year. Five years of that is about $106,000, kept.

So the trade is: keep $106,000 in your mid-fifties, forgo $59,300 dribbled out between 67 and 87. Even with no investment return at all on the money kept, they come out $46,600 ahead. Invest the difference at anything resembling market returns and it isn't close. To break even on those five years of contributions, they'd need to collect until past age 100. This person keeps their 40 credits, keeps premium-free Medicare, and their benefit stays north of $3,600 a month. The system's bend points did the work: by year 30 of a high-earning career, every additional dollar in buys benefits at the 15% rate on the way out.

I don't have to model this one. My wife and I have run this structure for about eleven years, dividing the work and the income between us, each drawing a salary at the FEIE maximum as it climbed from $100,800 in 2015 to $130,000 in 2025. Self-employment tax on those two salary streams would have been about $172,000 apiece: roughly $345,000 the household kept. Everything the business earned above the caps came home as S corporation distributions, which carry no SE tax in any scenario. The zeros went into two earnings records that already told very different stories.

Now run the same formula for a 32-year-old with eight years of credits, and the answer inverts. She hasn't hit the 40-credit floor that makes her eligible for any retirement benefit at all. Her early dollars land in the 90% bracket, the most valuable contribution dollars that exist. And she'd be giving up more than retirement: Social Security disability has a recent-work test, 20 of the last 40 quarters from age 31 on, so SSDI coverage quietly expires about five years after your last contribution no matter how long you paid in before. If you opt out young, price private disability insurance into the comparison, because you're dropping the public version.

Medicare has its own arithmetic. Forty quarters of coverage buys premium-free Part A at 65. Come up short and Part A costs $311 a month with 30 to 39 quarters, $565 a month with fewer than 30, at 2026 rates, for the rest of your life. If you're within a few quarters of 40, opting out early is an expensive rounding error.

Two people, one household, two different bets

The income split in my house wasn't a tax strategy; we divided the work and the pay followed it. But the same decision landed on two very different earnings records. I came into this with decades of contributions behind me. My wife had fewer than ten years, short of the forty credits that produce any retirement benefit at all.

For me the math was a rout, as the numbers above show. Her side was the close call, and it turns on a rule most couples don't know: a spouse can claim a benefit worth up to half of the other spouse's full retirement amount once the earning spouse files, and premium-free Medicare Part A on the other spouse's record, without a single credit of her own. Had she stayed in the system these eleven years, contributing on FEIE-cap earnings, her own record would eventually pay something like $2,300 a month. The spousal benefit she gets anyway is around $1,950. The difference, roughly $355 a month, adds up to about $85,000 over a 20-year retirement, against the $172,000 she kept by opting out. She comes out ahead before counting a dollar of investment return on the money kept. And survivorship quietly tips the scale further, because if I die first she steps up to 100% of my benefit in place of her own, regardless of what her own record says.

The lesson isn't that couples should both opt out. It's that this decision is made one earnings record at a time, and the spouse with the shorter record faces the harder math, in both directions.

Zoom out once more, because SE tax is only one line on the ledger. Stack the FEIE on top of the payroll-tax opt-out and a two-earner household at the caps keeps roughly $60,000 a year that the stateside version of the same couple sends to Washington: income tax excused by §911, payroll tax that never comes due under §3121(b). Invest that stream instead of remitting it and eleven years compounds past a million dollars: roughly $660,000 of tax never paid, plus what the market did with it while it waited, while the benefits you accrued in your earlier working life sit largely intact. One asterisk on the investing: income you exclude under the FEIE isn't compensation for IRA purposes, and a Seychelles shelf company sponsors no 401(k), so the compounding happens in taxable accounts. That's the real stake on the table. The corporation holding it open cost $800 a year.

That's the frame for the whole decision. Legality isn't the question; the statute answers that in plain text, and the compliance burden is the price of the answer being ordinary. The question is what your next contribution dollar buys you, and the Social Security Administration will show you that number for free. Mine was small enough to build a corporation around.