The Airport Effect

There’s a particular kind of anxiety that comes from not knowing how bad the results of a mistake can be. Rory Sutherland calls this ’the airport effect’. A thirty-minute delay is fine, unless it means you miss your connection, which means you’re sleeping in a terminal in Madrid. A carry-on bag is slightly overweight, so now you need to pay $150 and wait at baggage claim, and risk a lost bag. Your flight is delayed, and you don’t know if you should book another flight or wait. Trivial issues made stressful because you can’t tell whether it’s really trivial or catastrophic until the cascade either happens or doesn’t. You’re reacting to the uncertainty about whether the event matters.

I’ve noticed the same dynamic in financial life. When you can’t see the bottom, you make decisions defensively. The absence of a known floor turns every loss into a potentially existential one.

I have a spreadsheet with 106 columns that tracks every dollar I’ve earned, spent, saved, and invested since 2016. It projects my net worth to age 100 under various assumptions about market returns, inflation, and when I buy another car. I don’t plan to retire. But I wanted to know if I could, because knowing the life vest is there lets you stop being worried about standing on the deck.

Twenty-Five Times Expenses Is Too Low

The arithmetic of early retirement is advertised as simple. Take your annual spending and multiply by 25. If your portfolio exceeds that number, you can theoretically withdraw 4% per year, adjusted for inflation, and not run out of money over a 30-year retirement.

The “4% rule” is from Bengen’s 1994 research testing every 30-year period in US market history. It historically never failed.

I don’t need 30 years. I might need 50+, and exponential math gets noticeably less friendly when you extend the years. But the real issue is that the 4% rule was designed for a 65-year-old with a pension supplement, Medicare, and penalty-free access to every dollar they own. I have over a decade before I can touch my retirement accounts without a penalty, two kids hitting college within five years, and a spouse whose small business creates taxes I didn’t think about.

The Tyranny of the Withdrawal Equation

In 1897, Konstantin Tsiolkovsky derived the equation that governs all rocket propulsion. The math is referred to as “the tyranny of the rocket equation.” The fuel mass required to achieve a given velocity change grows exponentially, because the rocket must carry fuel to accelerate the fuel. You want to go a little faster? You need a lot more fuel. But the extra fuel is heavy, so you need even more fuel to carry that. The payload that reached the Moon was 1.5% of the launch mass.

Early retirement math has its own tyranny, small changes to assumptions have exponential impacts on results.

When you withdraw money from a 401(k) or IRA before age 59½, the IRS charges you ordinary income tax plus a 10% early withdrawal penalty. If your marginal federal rate is 22%, you keep 68 cents of every dollar you pull out. To net enough to cover a year of living expenses, you need to gross up: divide by 0.68. But here’s the recursive part: the grossed-up withdrawal is itself taxable income. A larger withdrawal pushes you into a higher bracket, which means you need even more, which pushes you higher still.

My spreadsheet modeled taxes as a flat percentage of withdrawals. Taxes are actually iterative. The actual cost of early withdrawals, once you let the feedback loop converge, is substantially higher than a flat-rate estimate suggests. Over nine years of pre-59½ withdrawals, the penalties alone add up to roughly 1.2× my annual spending.

The way rockets solved the tyranny of exponential fuel requirements was staging: burn a stage, drop the empty tank, continue lighter. Each stage is more efficient because it’s not carrying dead weight.

The FIRE community’s answer to the early-withdrawal penalty is the Roth conversion ladder. It’s the rocket staging of retirement planning.

Each year, you convert a chunk of money from your tax-deferred 401(k) or IRA into a Roth IRA. You pay ordinary income tax on the conversion but during early retirement, your income is low, so you’re converting in the 10% or 12% bracket instead of the 22-24% bracket you occupied while working. After five tax years, the converted amount can be withdrawn from the Roth penalty-free, at any age. Do this annually and you build a rolling stream of accessible funds: year one’s conversion unlocks in year six, year two’s in year seven, and so on.

The strategy has a well-known weakness, which is the five-year gap before the first rung matures. During years one through five, you need to fund your life from other sources — typically a taxable brokerage account, prior Roth contributions, or cash. This is called the “bridge.” Everyone who writes about Roth ladders talks about the bridge.

My bridge has a hole in it. In my theoretical case, the brokerage account (the bridge) runs out about a year before the first Roth conversions mature. The brokerage needs to fund 4-5 years of living expenses plus taxes, and with market growth it almost makes it. There’s a one-year gap where nothing is accessible without penalty.

And there’s a second constraint that the guides mention but I doing model: the Affordable Care Act subsidy cliff. If you buy health insurance on the ACA marketplace (as almost every early retiree must, for the decade-plus before Medicare), your premium subsidy depends on your Modified Adjusted Gross Income (MAGI). But Roth conversion income counts toward MAGI. Every dollar you convert to build your penalty-free ladder is also a dollar that pushes you closer to losing your healthcare subsidy.

In 2026, the ACA’s enhanced subsidies expired and the original subsidy cliff returned. If your household income exceeds 400% of the Federal Poverty Level (roughly the equivalent of 1.4× the median household income, for a family of four) you lose all premium tax credits. It is not ‘phased out’…$1 over the limit loses all of subsidies. For a couple in their 50s, the difference between subsidized and unsubsidized insurance is something like +15-25% of annual spending.

You’re trying to fill two different buckets (future tax-free withdrawals and current subsidized healthcare) from the same finite stream of allowable income. My spreadsheet had a column for Roth conversions and a column for healthcare costs but not anything connecting them.

The Three-Year Crunch

I went looking for a FIRE calculator that models the collision of early retirement with the ROTH ladder, ACA, and college-age children. I didn’t find one that actually worked. The person who can retire in their mid-40s statistically has kids between 8 and 16… so this seems like a meaningful omission. The college spending bomb arrives within five years of the earliest possible retirement date.

In my case: two kids, staggered by two years. Each is estimated to cost roughly a quarter of my annual spending per year of college. For six overlapping years, my effective spending jumps 50-100% above baseline. And this surge arrives during the exact window when the brokerage bridge is depleting, the Roth conversions haven’t matured, and every source of funds except the penalty-locked retirement accounts has been exhausted.

During those years, my revised model has no choice but to pull from the 401(k) with the 10% penalty. The tax feedback loop from above kicks in at full force. The ACA cliff gets blown, MAGI spikes to several multiples of the threshold, so you’re paying full-price health insurance on top of tuition. And because the universe enjoys ironic consistency, my car is estimated to need replacing in one of the peak years.

I suspect many people who run FIRE calculations with children simply don’t model the college years in detail. They assume scholarships, or 529 plans, or that the spending will somehow fit. FIRE calculators don’t model the interaction between college, penalty-locked accounts, the ACA cliff, and the Roth ladder timing. The compound effect of all four hitting simultaneously is not discussed at all, as far as I can tell.

This is exactly the kind of thing that generates the airport-delay anxiety when left unexamined. You know college is expensive, and retirement accounts are still locked behind penalties. But until you model the specific interaction, you don’t know if it’s a thirty-minute delay or a night in the terminal. You just know it’s probably bad, which is exactly the ambiguity that makes every subsequent decision feel fragile. Running the model turns “probably bad” into “$X over Y years, funded by Z”.

Four Differences

At this point I had two models of the same retirement: my original spreadsheet, which said I was fine, and the stress-tested version, which said I’d run out of money at 64 if O lost my job today.

It took embarrassingly long to find the discrepancies…

The first difference was spending. My spreadsheet tracks actual historical spending, and the categories it projects forward reflected what I’d been spending. The estimated spending was about 10% higher than expected, focused on two discretionary categories. A 10% difference in spending, at a 5%+ withdrawal rate, is the margin between a portfolio that slowly compounds forever and one that runs dry.

The second difference was taxes. My wife runs a small business and we file jointly. The original spreadsheet modeled my finances in isolation. The stress-tested version modeled the full joint return: her self-employment tax, her share of the federal brackets, the interaction between her income and our household MAGI. Fixing this moved the model from “fails at 64” to “works into the 80s.”

The third difference was half a percentage point of investment returns. My original model used roughly 7.5% nominal return (before subtracting cost inflation), but 7% was better supported by research. Small difference, but over 40 years, it’s the difference between the money lasting to 74 or to 84. A half-point of annual return was worth a decade of retirement.

The fourth was just an obvious mistake. When the portfolio depleted and there was nothing left to withdraw, my spreadsheet didn’t flag an error or throw a warning. It just… stopped withdrawing. The accounts hit zero and the model continued forward, showing a small remaining balance that was actually just the HSA growing quietly in the corner.

The Average Return Is Irrelevant

Even after correcting all four discrepancies the model reveals something that no amount of tax optimization can fix.

Social Security at 70 covers less than half of spending. The gap between benefit and expenses is permanent, and it grows slowly because expenses inflate at 3% while the benefit may not.

The standard FIRE analysis handles this by assuming the portfolio compounds indefinitely. Withdraw 4%, earn 7%, the difference grows, everything’s fine. But Morningstar’s “State of Retirement Income” study (the most rigorous ongoing analysis of this question I’ve found) shows what actually determines outcomes: not the long-run average return, but the returns during the first five to ten years of withdrawal.

Morningstar found that retirees who avoided investment losses during the first five years were dramatically less likely to exhaust their savings. Negative returns during that early window account for roughly 70% of retirement plan failures. The mechanism is straightforward: losses while withdrawals are large create a compounding deficit that later gains can’t overcome, because the base has been permanently reduced. Two people with identical portfolios, identical spending, and identical 20-year average returns can end up half a million dollars apart based purely on which years the losses fell in. Your average return is nearly irrelevant. What matters is whether the market cooperates during the specific window when your portfolio is large and your withdrawals are large and you haven’t yet built the cushion that makes later downturns survivable.

This is sobering. But the analysis also surfaced a backstop that most retirement models treat as scenery: the house.

Vanguard pointed out that incorporating home equity into the retirement math increases retirement readiness dramatically. For most people, including us, the house is the largest asset, and the conventional wisdom is to leave it alone. It’s where you live and it’s not liquid. But the conventional wisdom was written for 65-year-olds with pensions, not people running 55-year projections.

Wade Pfau’s research highlights one counterintuitive tool: a reverse mortgage line of credit opened at 62, left untouched, grows at the loan interest rate more than doubling by 80. You never touch it unless you need it, and if you do, you’re drawing liquidity instead of selling depressed equities during a downturn.

In the model, when I let the home equity serve as the late-retirement backstop (through a combination of downsizing when the kids leave and a standby HECM on the smaller home) the plan survives comfortably into the late 80s at conservative assumptions and past 90 at moderate ones.

Marked Hazards

I’m not retiring. I like what I do, I’m good at it, and the marginal year of work currently buys more optionality than the marginal year of not-work. But the model still helps.

Before the analysis, I had a vague confidence that my family was “probably fine”… but that confidence dissolves at 2 AM when you read about funding cuts or notice that your 401(k) dropped 8% in a month.

After the analysis, I’m sure the model is still wrong, just less wrong than it was. But what I have is a better map of the failure modes.

Income uncertainty can trigger “what if this escalates” anxiety because of unmapped consequences. Now that I have a map, the consequences are: if everything goes wrong at once (job loss, bad market, no new income) I have a documented path to a reduced but functional life funded by a combination of tax-efficient withdrawals, healthcare subsidy management, Social Security timing, and a house that is useful as a financial instrument.

The model also has real vulnerabilities; a severe market crash in the first five years being the biggest. But there’s a meaningful difference between “I don’t know what would happen” and “I know what would happen and here’s how I’d manage it.”

I think more people should do this exercise, not because they should retire early, but because the ambient anxiety of unexamined risk is a tax on every decision you make. You optimize for local safety instead of global optionality. A spreadsheet can be a map with marked hazards, and navigating marked hazards is a qualitatively different experience than navigating blind.