Every retirement account is a deal with the tax collector: shelter your money by their rules, and they take a smaller cut — or take it at a friendlier time. This guide works through the whole bargain, one module at a time: where these accounts came from, how the tax math actually works, what employers match and why, every legal door out before retirement age, and the strategy that falls out of all of it.
One module is written at a time, and the doc grows as Noah works through it. Bold rules mean the module is live; faded rows are planned but not yet taught.
How America went from pensions it never had to think about, to accounts it has to drive itself — and why every rule you'll meet later in this guide (the match, the penalty, the required withdrawals) is a scar from this history.
Because IRAs and 401(k)s are not products anyone designed from scratch. They are accumulated legislation — layers of tax law passed in different decades to patch different problems. If you memorize the rules as arbitrary trivia ("you can take $10,000 for a first home from an IRA but not a 401(k)"), they won't stick and they'll constantly surprise you. If you know which problem each layer was patching, the rules become almost predictable. Every module that follows leans on this one.
So the single concept for this module is: the shift from defined-benefit to defined-contribution retirement — what those words mean, why the shift happened, and what it moved onto your shoulders.
For most of the twentieth century, American retirement planning was described as a stool with three legs. You personally were only responsible for one of them.
A government-run floor, created in the depths of the Depression. Funded by payroll taxes, paid as a lifetime annuity. You never manage it; it simply arrives.
A defined-benefit plan: the company promises a formula — say, 1.5% of final salary per year of service — payable for life. The company invests the money, bears the risk, and does the math.
Whatever you put in the bank or the market yourself — fully taxed, no special treatment, and for most households, the smallest leg by far.
Notice the word defined-benefit. The thing being guaranteed is the benefit — the monthly check. How the employer gets there is the employer's problem. If markets crash, the company owes you the same check. If you live to 104, the company keeps paying. The employee's exposure to investment returns and to their own longevity is, in theory, zero.
In theory. The catch is hiding in that phrase "the company promises."
Pension promises were only as good as the company making them — and for decades, almost nothing in federal law forced companies to actually set aside enough money to keep them. The failure that made this impossible to ignore came when the Studebaker auto company shut its South Bend, Indiana plant in December 1963. Its pension plan was badly underfunded: workers 60 and older got their pensions, thousands of vested workers in their 40s and 50s got roughly fifteen cents on the dollar, and everyone else got nothing.
It took a decade of hearings, but the answer arrived on Labor Day 1974.
ERISA, 1974 — and the accidental birth of the IRA
The Employee Retirement Income Security Act did three big things: it set minimum funding and vesting standards for pensions, it created the PBGC (a federal insurer that pays pensions when a plan fails), and — almost as a footnote — it created the Individual Retirement Account.
The original IRA was not for everyone. It was for workers without an employer plan: a portable, self-owned substitute, capped at $1,500 a year (or 15% of pay). The core design — tax-deductible going in, tax-deferred while growing, taxed coming out, with a 10% penalty before age 59½ — is the DNA that every later account inherits. That penalty, which Module VI is entirely about, is as old as the account itself: Congress gives you the tax break because the money is fenced for retirement, and the penalty is the fence.
Here is the strangest fact in this whole subject: nobody designed the 401(k) to be the nation's retirement system. The Revenue Act of 1978 added a short, obscure subsection — section 401, paragraph (k) — to the tax code, effective 1980. Its intent was narrow: to settle a long-running dispute about executives deferring bonuses, by saying an employee wouldn't be taxed on compensation they chose to defer into a plan.
In 1980, a Pennsylvania benefits consultant named Ted Benna read the paragraph and saw something no one intended: it could support a plan where ordinary employees divert part of every paycheck pre-tax, and — his real innovation — where the employer adds a matching contribution as the carrot to get people to participate. He built the first such plan for his own firm, the Johnson Companies. In 1981 the IRS blessed the salary-reduction interpretation, and corporate America moved with astonishing speed: for employers, a match you only pay when workers opt in is vastly cheaper and more predictable than a lifetime pension promise.
The same year, 1981, Congress opened the IRA to all workers (raising the cap to $2,000), turning it from a pension substitute into a universal sidecar. The modern two-account world — employer plan plus individual account — was in place by 1982.
Nine dates that built the system
Step through the whole legislative history — each stop is one law and the problem it was patching. The later modules of this guide hang off these dates.
The pension world and the 401(k) world differ in one fundamental way, and it isn't the paperwork. It's who carries the risk. A defined-benefit plan defines the output (your check) and leaves the inputs to the employer. A defined-contribution plan defines only the input (what goes in each payday) — the output is whatever your investments turn it into, over however long you happen to live.
Who carries it?
Flip between the two worlds. Neither is strictly better: the pension's guarantees came bundled with immobility and dependence on one company's solvency; the 401(k)'s freedom came bundled with two risks most people are poorly equipped to price.
The migration was fast and one-directional. Among private-sector workers who have a workplace plan, the share relying on a traditional pension collapsed within a generation — and most surviving pensions are closed to new workers ("frozen"). Figures below are approximate shares of active private-sector plan participants whose primary plan is defined-benefit, from Department of Labor data.
Pensions as the primary plan, private sector
Public-sector workers (teachers, police, federal employees) are the big exception — defined-benefit pensions still dominate there, which is why this guide's rules mostly won't apply to friends with government jobs.
"The 401(k) was designed as a complete retirement system, so if mine isn't enough, I'm using it wrong." False in both halves. It was an accounting footnote repurposed by one consultant; it was originally imagined as a supplement sitting on top of pensions and Social Security, not a replacement for them. Even Ted Benna has publicly lamented how complex and load-bearing his creation became.
The practical consequence: the system genuinely does leave gaps — it works best for steady earners at employers with good matches, and worst for gig workers, job-hoppers who cash out, and low earners with no plan at all. Congress keeps bolting on patches (auto-enrollment in 2006, the SECURE Acts in 2019 and 2022) precisely because the foundation was never engineered for the weight it carries.
The employer match (Module III) exists because Benna needed a carrot. The 10% penalty and its exceptions (Modules VI–VII) exist because sheltered money needs a fence, and Congress keeps cutting sympathetic gates into it — medical bills, first homes, new babies, disasters.
Required minimum distributions (Module VIII) exist because the deal was tax deferral, not tax escape — the IRS eventually wants its cut. Hold onto "it's a deal, with a fence, and gates" — it organizes everything ahead.
Now that you know the accounts are deals with the tax collector, the obvious next question is: what exactly is the deal worth? "Tax-deferred" and "tax-free" get used loosely, and the traditional-versus-Roth choice — pay tax now or pay tax later — is the single highest-stakes decision in this whole subject. Module II builds the tax math from scratch, with a calculator you can push on until the logic is obvious.
Answer in your head first, then reveal. If either of the first two feels shaky, reread the matching section — Module II assumes them.
Workers without an employer pension. After Studebaker showed that pension promises could evaporate, ERISA regulated the promises — and the IRA was the portable, self-owned fallback for people who had no promise at all. It only became a universal account in 1981.
Section 401(k) was written in 1978 to settle a narrow dispute about deferred executive compensation. Ted Benna reinterpreted it in 1980 as a mass-market salary-reduction plan with an employer match. Congress never sat down and designed a national retirement vehicle — one consultant found a loophole-shaped door and the entire economy walked through it.
Investment risk (your balance is whatever markets make of your contributions) and longevity risk (a pension pays until you die; a 401(k) balance can run out). Portability moved the other way — that's the trade.
Three doors for the same dollar — taxable, traditional, Roth. What "tax-advantaged" actually buys you, the one-line algebra that makes the traditional-vs-Roth choice honest, and the bracket asymmetry that most people never notice.
Module I ended with a frame: every retirement account is a deal, with a fence, and gates. This module is about the deal itself. Before you can care about matches, limits, or penalty exceptions, you need to answer the plain question: what is the tax advantage actually worth, and which flavor of it should you take? "Tax-deferred" and "tax-free" get thrown around as if they were synonyms. They are not, and the difference is the single highest-stakes recurring decision in this whole subject.
The concept for this module: a dollar of income can reach your retirement through three doors, and the doors differ only in when the IRS takes its cut.
Follow one dollar of salary through four stages: you earn it, you contribute it, it grows for decades, you withdraw it. Each door taxes different stages. That's the entire taxonomy — every account in Modules III and IV is just one of these doors wearing a uniform.
Earn: taxed as income.
Grow: dividends taxed every year; gains taxed when you sell — though at the gentler long-term capital-gains rates.
Withdraw: no extra tax, no rules, no fence.
Taxed twice, but lightly the second time — and totally unrestricted. This is the door the other two are measured against.
Earn: not taxed — the contribution is deducted.
Grow: untaxed, entirely.
Withdraw: every dollar taxed as ordinary income, at whatever your rate is then.
This is the 1974 IRA's DNA: deferral. The IRS isn't waiving its cut — it's waiting, and it will insist (that's Module VIII's required distributions).
Earn: taxed as normal income.
Grow: untaxed, entirely.
Withdraw: zero tax, forever, if the withdrawal is qualified.
The 1997 invention. Pay once at the door and the IRS never sees that money again — no tax on decades of growth, and (for Roth IRAs) no required distributions either.
Here is the most clarifying — and most widely missed — fact in retirement planning. Suppose your tax rate is 25% now and 25% at withdrawal, you have $10,000 of pre-tax salary, and your investments will multiply by 8 over the decades. Watch both doors:
Identical. Not close — identical, because multiplication commutes: contribution × growth × (1 − tax) gives the same answer no matter which order you apply the terms. The Roth's celebrated "tax-free growth" and the traditional's celebrated "upfront deduction" are, at equal tax rates, the same advantage viewed from opposite ends.
This collapses the whole decision into one question: is your tax rate higher now, or at withdrawal? Higher now → traditional wins (deduct at the high rate, pay at the low one). Higher later → Roth wins. Equal → tie. Everything else is noise on top of this identity. Push on it yourself:
One salary dollar, three doors
Drag the slidersStart with $10,000 of pre-tax salary and send it through each door. The taxable door pays income tax at entry and a 15% long-term capital-gains tax on growth at exit (simplified: no annual dividend drag).
Full amount in, taxed as ordinary income on the way out.
Taxed at entry, then untouchable forever.
Taxed at entry, growth taxed 15% at exit.
So the decision is a bet on now versus later. But here's the subtlety that tilts the bet for many people, and it comes from how brackets work. Your traditional contribution is deducted off the top of your income — it all escapes tax at your highest, marginal rate. Your retirement withdrawals, if they're your main income, fill the brackets from the bottom: the first slice is covered by the standard deduction at 0%, the next slice at 10%, then 12%, and so on. The average (effective) rate on a withdrawal is always lower than the top bracket it reaches.
So even if you retire into the same bracket you worked in, traditional dollars often win: they skipped tax at the top rate and re-enter at a blended rate. Drag the withdrawal slider and watch the blend:
Withdrawals fill the brackets from the bottom
Illustrative single-filer brackets, rounded24%
What a working-years traditional contribution escaped, off the top.
The rate comparison is the engine of the decision, but four real-world quirks lean on the scale. These matter most when the rate bet is close to a coin flip:
Limits are written in nominal dollars — the same cap for both doors. But a capped Roth contribution is after-tax money: it represents more pre-tax salary, so a maxed-out Roth shelters strictly more purchasing power than a maxed-out traditional.
Roth IRAs are exempt from the required minimum distributions of Module VIII — the money can compound untaxed for life and pass to heirs. Traditional balances get forcibly drained (and taxed) on the IRS's schedule.
You can convert traditional money to Roth later — deliberately, in low-income years (a gap year, early retirement), paying tax at bargain rates. There is no reverse move: Roth can never become traditional. Deferring keeps the option alive.
Long-term gains rates are gentle (0% for modest incomes), there's no fence and no gates, and heirs receive a stepped-up basis. Once the sheltered doors are maxed, the taxable door is the normal next stop — Module X sequences all three.
"Roth is better because the growth is tax-free." The worked example above shows why this is wrong as stated: at equal rates, the traditional's upfront deduction manufactures exactly as much extra wealth as the Roth's tax-free growth. People fall for it because $80,000-taxed-to-$60,000 feels worse than $60,000-untouched — but they're the same number. The growth story only becomes a real argument via the nominal-cap quirk in the tile above.
The mirror-image error — "I'll obviously be in a lower bracket when retired, so traditional always wins" — is more often true but still a bet, not a law: required distributions on a large balance, a pension, Social Security, a working spouse, or simply Congress raising rates can push your later rate up. That's why the standard advice for people unsure about their trajectory is to hold both flavors — "tax diversification" — and choose which to draw from year by year.
Early career / low bracket: Roth — you're paying the entry tax at the cheapest rate of your life.
Peak earnings: traditional — deduct at your top marginal rate, re-enter at a blended one.
Genuinely unsure: split. Owning both doors is a hedge that also buys year-by-year control of your retirement tax bill.
You now hold the core deal in your hands: shelter from the growth-stage tax, paid for at entry (Roth) or exit (traditional). But nobody meets these deals in the abstract — they meet them wrapped inside an employer plan, where a second force appears that dwarfs the tax math entirely: the match, Ted Benna's carrot from Module I, an instant 50–100% return that no bracket arithmetic can compete with. Module III opens the 401(k) machine: deferrals, match formulas, vesting schedules, and the plan's own limits.
These three carry the module. The first one, especially — Modules V and X quietly assume it.
Because the final wealth is contribution × growth × (1 − tax rate), and multiplication doesn't care about order. Traditional applies the tax factor last; Roth applies it first; the product is the same. The decision therefore reduces to comparing the rate now versus the rate at withdrawal — nothing else is in the formula.
Contributions come off the top of a salary that has already filled the lower brackets — every deducted dollar escapes your highest rate. Withdrawals that constitute your main retirement income fill the brackets from the bottom — standard deduction first at 0%, then 10%, 12%, and up — so their average rate is lower than the top bracket they touch. It favors traditional, and it's why "same bracket in retirement" still usually means "lower rate in retirement."
Any two of: a maxed-out Roth shelters more real purchasing power (the nominal cap is after-tax dollars); Roth IRAs have no required minimum distributions; qualified withdrawals can't be pushed higher by future tax-rate increases; and — a preview of Module VII — Roth contributions can come back out at any age, tax- and penalty-free.