The New American Accord ends tax deferral — everywhere, for everyone, with no exceptions. No 401(k), no IRA, no Roth for new money, no deferred executive compensation, no insurance policy with an investment account inside, no ETF conversion, no synthetic holding, no small-saver carve-out. Gains are taxed when taken. Income is taxed when earned. Where the law still allows a delay — an installment sale, an illiquid stake — the delay carries interest. This is the largest single simplification in the Accord and, after healthcare, its largest recapture of hidden subsidy. Here is the full accounting: why, what moves, what shrinks, and what it recovers.
What deferral actually was
Deferral is a loan. The Treasury lends you your own tax bill, interest-free, for as long as the gain stays unrealized. Congress never enacted this as a loan program, never set terms, never means-tested it — yet it grew into one of the largest expenditures in the federal budget, on one underwriting rule: the loan scaled with sophistication. A worker with no tax liability got nothing. A median saver got a modest advance. A founder with counsel got decades of compounding on the Treasury's money and, until the Accord eliminated basis step-up, forgiveness of the whole loan at death.
Historical grounding
Deferral began as an administrative compromise. In 1921, when Congress first allowed nonrecognition exchanges, no one could track cost basis across decades of paper records; "settle up later" was a concession to the ledger technology of the time. Pensions in the 1940s, the IRA in 1974, and the 401(k) in 1978 inherited that logic and added a second claim: deferral would encourage saving.
Then the engineering started, and its record is consistent. Congress capped the IRA; the industry produced the backdoor Roth. Congress banned the swap fund by name in 1966; the industry spent sixty years building a compliant replacement and launched it as the 351 ETF conversion — 105 funds, $22 billion in assets, $6.5 billion in deferred gains as of this July. Every specific fence became a design target. The general lesson is the Accord's Preference Principle: any tax preference at scale gets captured, and timing preferences are the easiest to hide.
Meanwhile the original excuse expired. The same public rail that computes withholding now computes realized gains continuously and files the results automatically. Annual honest taxation was impractical in 1954. It is trivial now.
Personal grounding
A nurse who sells one stock and buys another in a brokerage account pays tax that afternoon. She always has. A family that moved a half-billion-dollar position into a custom ETF paid nothing. The Accord extends the nurse's rule to everyone. The rail does the paperwork; no one files anything.
A second personal fact is rarely stated: a traditional 401(k) was never entirely the saver's. Roughly a fifth to a quarter of every balance was the government's deferred tax claim, embedded in the statement. The Accord's account is post-tax and settled annually, so the balance on the screen is simply the saver's money — no required distributions, no penalty ages, no withdrawal rules. Ownership becomes clearer, not weaker. What ends is the commingling.
Ethical grounding
Three principles. Equal treatment: identical transactions should bear identical tax, whatever wrapper performs them. Honest subsidy: help delivered through a tax bracket is help proportional to income — most of the retirement expenditure's value went to the top fifth of earners and nearly none below the median, a design no Congress would enact as visible spending. Universal participation: the Accord rejected even a sympathetic small-saver exemption, because any zero-tax class becomes a seam that widens. Everyone pays into the base. Support comes from the floor — the universal guarantees — not from exemptions in the code.
Policy grounding
The decisive evidence is the Danish registry study: millions of savers over four decades. About 85 percent of people do not respond to tax subsidies at all; each subsidy dollar produced roughly one cent of new saving. What moves saving is the default. Automatic enrollment raised participation from 37 to 86 percent in the landmark U.S. study; the U.K.'s national auto-enrollment holds opt-out near nine percent.
So the Accord keeps what works and stops paying for what doesn't. Every worker is enrolled by default at 8 percent of pay into the public account — close to the paycheck share freed when the health premium disappears, so take-home pay is roughly unchanged. Anyone can adjust or stop it in one step, and withdraw at any time without penalty. Because the account is a single lifetime account, the leakage that weakened defaults in every prior system — cashing out the old plan at each job change — cannot occur. There is no old plan to cash out.
What moves: the hidden tax claim inside retirement accounts
American retirement accounts hold roughly $40–45 trillion, about $30 trillion of it tax-deferred. Inside those balances sits the embedded federal claim — the deferred tax — of roughly $6–8 trillion. Ending deferral confiscates nothing: existing balances run off under current law exactly as promised. What changes is that no new deferral is created in front of the claim, and every new dollar of gain in the country settles with the ledger annually. After the transition, personal accounts are fully personal, the tax claim is collected as it accrues, and the ambiguity that supported a planning industry is gone.
What shrinks
| Extractive layer | Approximate current scale | Under the Accord |
|---|---|---|
| Retirement tax expenditure (deferral + inside buildup) | ~$300–400B/yr | Recaptured as flows shift and balances run off |
| Asset-management and plan fees above institutional cost | ~$100–200B/yr | Compressed toward the 0.03–0.05% public benchmark |
| Retirement plumbing: recordkeeping, administration, compliance testing | Tens of billions/yr, hundreds of code pages | Category deleted — no limits, no testing, no distribution policing |
| Deferral structuring: wrappers, conversions, the planning bar | Fees on $20B+/yr of conversions alone | No free timing exists to sell |
| Deferral's share of the $530B/yr filing burden | A meaningful fraction of household complexity | Rail computes and withholds; citizens file nothing |
Each row is a workforce contraction the Accord causes, and it applies its standard transition support: wage insurance — a temporary top-up for workers who take a lower-paying job while retraining — and first claim on Skills Wallet funds, the lifetime training account, for certification and coursework. Advisory work that clients choose to buy survives on its merits.
What it recovers from the top
Most of the retirement expenditure went to the top quintile — on the order of $180–250 billion a year now redirected to the public ledger. Add annual taxation of fund gains that wrapper mechanics previously erased, tax at every asset switch that previously passed through a nonrecognition door, interest on installment and synthetic delays, and settlement at death, and the recovery is concentrated where the loans were largest. All magnitudes are pending the published scoring, with one methodological note: past estimates of how the wealthy respond to capital-gains taxes were measured in a code full of exits. This code has none, and death settles the remainder.
Other observations
Equity markets reprice once: the previously untaxed pension pool and the end of lock-in compress valuations at transition. Stated plainly, lower prices for today's holders are higher returns for young savers just starting to buy. The simplification is verifiable: if anyone finds a remaining path to free deferral, the disclosure law pays a bounty for reporting it. And the deepest change is a habit: a code with no timing games retires the assumption that the published rate is an opening bid.
The Accord taxes income when earned, consumption when spent, harm when caused, wealth when transferred, and gains when taken, and it charges interest on any delay the law still allows. The loan program is closed. What replaces it shows up in the paycheck: a default that works, an account with near-zero fees, a floor that holds, and a balance that is entirely yours.
Sources and flags (verify all on publish): retirement asset totals ~$40–45T, deferred share ~$30T (ICI); retirement tax expenditure $250–400B/yr by measure (JCT/Treasury — reconcile basis); top-quintile share (CBO distributional series); Chetty et al., QJE 2014 (~85% passive; ~$0.01 new saving per subsidy dollar); Madrian–Shea 2001 (37%→86%); UK NEST opt-out ~9%; embedded claim ($6–8T) = deferred balances × 20–25% effective rate, assumptions stated; fee excess = deferred AUM × (average fee − institutional benchmark), assumptions stated; 351 conversions (Bloomberg, July 2026); all recapture magnitudes WORKBOOK-PENDING before quoting as scored.