[ER-05] Accessing Money Early

Ten sources, ranked — because before 59½ the question is not “how much can I take” but “how much income does taking it create”

Pillar: Early Retirement Mechanics · Applies to: Anyone spending from a portfolio before 59½, and anyone building the bridge that will let them Last verified: August 2026 · Refresh cadence: Evergreen (mechanics); figures via FN-02 Related: ER-02 Roth Conversion Ladder · ER-03 SEPP / 72(t) · ER-04 Rule of 55 · TR-01 Five-Year Runway · HC-01 ACA Bridge · HC-03 Medicaid Floor · HC-04 HSA Mastery · TX-03 Withdrawal Sequencing

Not advice. The ordering rules, five-year clocks, and plan-type distinctions here are unforgiving and interact badly with each other. A withdrawal that looks identical from two different accounts can differ by a 10% penalty and an entire year’s health subsidy. Confirm your own account types and clock start dates before spending anything.

---

TL;DR

• Before 59½ the binding constraint is rarely the penalty — it is MAGI, because MAGI sets your ACA subsidy (HC-01) and your Medicaid floor (HC-03). Rank your sources by income generated per dollar spent, not by tax rate.
• §3 funds an identical $80,000 of spending two ways. One produces $10,000 of MAGI; the other produces $80,000. Same spending, same household, different pockets.
• Roth IRA ordering rules are the most useful rule in early retirement: contributions come out first, always, tax- and penalty-free, at any age, regardless of how long the account has existed ✅. Then conversions, oldest first. Then earnings.
• A governmental 457(b) has no 10% early-distribution penalty at any age after separation ✅ — the most underrated bridge asset there is. Its non-governmental cousin is a different and riskier animal.
• The goal is to hit a MAGI target, not to minimise MAGI. §4: the same optimisation that dodges the 400% cliff can drop you under the 138% floor, where subsidies stop entirely.

---

1. Why MAGI is the ranking, not tax

Most early-access content ranks sources by tax cost. That is the right ranking for someone with employer health insurance and the wrong one for almost everyone this article serves.

For a pre-Medicare retiree buying coverage on the marketplace, a dollar of MAGI does three jobs at once: it is taxed, it shrinks the premium tax credit, and near a threshold it can vaporise one. The cliff at 400% FPL — $84,600 for a household of two ✅ (FN-02) — means the marginal cost of one dollar can be five figures (HC-01 §4). No tax table shows that.

So the useful question about each source is: how many dollars of MAGI does it cost me to spend one dollar? That ratio, not the rate, is what the ranking below is built on.

2. The ten sources, ranked

#	Source	MAGI per $1 spent	Penalty before 59½	The catch
1	Cash / money market	$0	none	Interest earned is still income; the principal is not
2	HSA reimbursement for old receipts	$0 ✅	none	Needs documented past medical expenses (HC-04)
3	Taxable basis (return of capital)	$0	none	Requires specific-ID lot selection to isolate (TX-01)
4	Roth IRA contributions	$0 ✅	none ✅	Spends your most flexible asset
5	Seasoned Roth conversions (5+ yrs)	$0 ✅	none ✅	The tax was paid five years ago (ER-02)
6	Taxable gains	only the gain	none	May fit the 0% LTCG bracket (TX-01)
7	Governmental 457(b)	$1	none, any age ✅	Only after separation; rolling to an IRA forfeits it
8	Rule of 55 plan withdrawals	$1	none ✅	Separation at 55+, that plan only (ER-04)
9	SEPP / 72(t)	$1	none, if unbroken	Locked schedule, retroactive bust penalty (ER-03)
10	Bare traditional withdrawal	$1	10%	The fallback that means the plan failed

Rows 1–5 are the reason a well-built bridge feels like magic: five ways to spend real money while reporting almost no income.

Roth IRA ordering rules

Worth stating exactly, because it is the single most useful mechanic here. The IRS treats all your Roth IRAs as one account ✅, and distributions come out in a fixed order ✅:

1. Contributions — always tax-free and penalty-free, at any age, regardless of account age ✅. No five-year wait applies to your own contributions.
2. Conversions, oldest first (FIFO) ✅ — tax already paid; penalty-free once the conversion is five years old or you are 59½ ✅. The “or” matters.
3. Earnings — last, and taxable plus penalised unless the distribution is qualified.

Two five-year clocks get confused constantly. One governs earnings being tax-free and starts with your first-ever Roth IRA contribution. The other governs each conversion’s penalty exemption and starts 1 January of that conversion’s tax year — every conversion carries its own ✅. A ladder is simply a queue of these clocks (ER-02).

Governmental 457(b): the underrated one

Distributions from a governmental 457(b) after separation are not subject to the 10% early-distribution tax, at any age ✅. No age-55 requirement, no schedule, no lock-in. For someone retiring at 48 with a public-sector career behind them, this is the whole bridge.

But the two kinds of 457(b) are not the same asset:

	Governmental	Non-governmental (tax-exempt / “top hat”)
Assets held	In trust for employees ✅	A general asset of the employer, exposed to its creditors ✅
Rollable to an IRA	Yes ✅	Generally no ✅
Early-withdrawal penalty	None after separation, any age ✅	None, but distribution timing is set by the plan ✅

The non-governmental version carries employer credit risk until it is paid out ✅ — if the employer fails, you are an unsecured creditor. That changes both how fast you should take it and how much of your net worth belongs there.

And the same trap as ER-04: rolling a governmental 457(b) into an IRA converts it into IRA money and forfeits the penalty exemption ◻️. Keep it where it is until you have spent what you need before 59½.

Statutory exceptions worth knowing

Beyond the ranked sources, § 72(t) waives the penalty for specific circumstances. SECURE 2.0 added several, and one goes live for the first time this year:

Exception	Cap	Notes
Disability; death	none	Long-standing ✅
Unreimbursed medical expenses	above the AGI floor	Long-standing ✅
Health insurance premiums while unemployed	premiums paid	IRA only ✅ — directly relevant to an ACA bridge
Higher education	qualified costs	IRA only ✅
First home	$10,000 lifetime	IRA only ✅
Birth or adoption	$5,000 per child ✅	Repayable ◻️
Qualified disaster recovery	$22,000, not indexed ✅	Federally declared disaster area (SECURE 2.0 § 331)
Domestic abuse victim	lesser of ~$10,000 (indexed) or 50% of the account ✅	SECURE 2.0 § 314
Emergency personal expense	$1,000 per calendar year, one distribution ✅	SECURE 2.0 § 115
Terminal illness	none ✅	Physician certification required ◻️
Long-term care insurance premiums	~$2,500/yr ◻️	SECURE 2.0 § 334 — first available for distributions in 2026 ◻️ (HC-08)

None of these is a plan; all are escape hatches. The one worth actually building around is the unemployed-health-premium exception, because it is uncapped and lands precisely where an early retiree’s costs do. Confirm current amounts against the IRS exceptions table — several are indexed and the newest are still accreting guidance.

Borrowing: the zero-MAGI source that is not income at all

A loan is not a distribution, so it creates no MAGI whatsoever — which makes a securities-backed line of credit, a HELOC, or margin the most MAGI-efficient money in existence. Two legitimate uses: bridging a lumpy expense across a tax year so a single large need does not blow a subsidy cliff, and avoiding a forced sale in a down market (ER-06).

The honest accounting: it is leverage against a portfolio you are already drawing down, the rate floats, and a margin call arrives exactly when markets have fallen — the same moment your other assets are worth least. Open the line while you still have W-2 income (TR-01 §3), size it small, and treat it as insurance rather than income. It belongs on this list for completeness, not near the top of it.

3. The same $80,000, two ways

A couple, both 57, spending $80,000 and buying marketplace coverage.

The managed stack:

Source	Spent	MAGI created
Cash / money market	$15,000	$0
HSA reimbursement, shoeboxed receipts	$8,000	$0 ✅
Taxable sale of $40,000, basis $30,000	$40,000	$10,000 — only the gain
Roth IRA contribution basis	$10,000	$0 ✅
Seasoned Roth conversion (5+ years old)	$7,000	$0 ✅
Total	$80,000	$10,000

The naive alternative: take the whole $80,000 from a traditional IRA. Spending: $80,000. MAGI: $80,000 — and a 10% penalty on all of it unless ER-04 or ER-03 applies.

Identical spending. MAGI of $10,000 or $80,000, decided entirely by which pocket it came out of. In FPL terms for a household of two ✅: 47% versus 378%.

The third row is a choice, not a fact

“Sell $40,000 of stock” does not have one MAGI answer. It has as many answers as you have lots, and the default setting picks the worst one:

Lots sold to raise $40,000	Basis	Gain = MAGI
High-basis lots (bought recently)	$36,000	$4,000
Blended / average	$30,000	$10,000
FIFO default — oldest, lowest-basis shares	$12,000	$28,000

Same sale, same $40,000 in your pocket, and a $24,000 spread in reported income. FIFO is the default at most brokers, and it sells your oldest and therefore usually lowest-basis shares first. Setting cost-basis method to specific identification — once, in the account settings, before you need it — is what converts row 3 from a fact into a lever (TX-01).

4. The trap in the optimisation: too low is also wrong

Read that result again and the flaw is obvious. 47% of FPL is below the 138% Medicaid floor ($29,187 for a household of two ✅) — in an expansion state that means Medicaid rather than a subsidised marketplace plan, and in a non-expansion state it means the coverage gap, where neither is available (HC-03). The optimisation succeeded and produced the wrong answer.

So the objective is a target, not a minimum. Say the target is $35,000 — 165% of FPL, above the 138% floor and below the 200% CSR line ($42,300 ✅), which buys a Silver plan at 87% actuarial value ✅ (HC-01 §5).

The stack above generates $10,000. The gap is $25,000 of MAGI that must be manufactured without spending more money. The lever is a Roth conversion: it adds MAGI, costs nothing in cash flow, and seasons for penalty-free use in five years (ER-02). Spending stays $80,000; MAGI rises to target; the household buys a 87%-AV Silver plan and banks a future ladder rung.

This is the central skill of the pre-Medicare years: spending and reported income are two dials, and they are not connected. TX-03 sequences them across the whole retirement; this page is the parts list.

5. Building the sources before you need them

Everything above is an inventory problem, and the inventory has to exist on day one (TR-01 §7).

• Roth contribution basis is the most valuable dollar you own — spendable at any age, penalty-free, MAGI-free, and it never expires. Spend it last among the zero-MAGI sources despite being the most tempting, because it is the only one that also solves emergencies.
• Keep the HSA receipts. There is no deadline to reimburse yourself for a qualified expense ✅, so every documented receipt is a future MAGI-free withdrawal (HC-04).
• Manage taxable basis deliberately. High-basis lots are cheap to spend; specific-ID lot selection is what lets you choose (TX-01).
• Start the ladder five years before you need its first rung (ER-02).
• Do not roll the 457(b) or the Rule of 55 plan until you have spent what you need from them.

Three Scenarios

A. The public-sector retiree at 48. She has a governmental 457(b) and a 403(b). The 457(b) has no penalty at any age ✅; the 403(b) does, and she is seven years from ER-04 eligibility. She spends the 457(b) first and leaves the 403(b) untouched — the reverse of the usual “spend taxable first” advice, and correct because the 457(b) is the only penalty-free pre-55 source she has. Rolling it to an IRA “to consolidate” would have cost her the exemption.

B. The 52-year-old with a large Roth and no plan. He has $400,000 in a Roth IRA, of which $180,000 is contribution basis. He believes he cannot touch it until 59½. He can withdraw the entire $180,000 today, tax-free, penalty-free, and MAGI-free ✅. The constraint he thought he had does not exist, and it changed his retirement date by four years.

C. The couple optimising themselves into the coverage gap. They engineered MAGI down to $12,000 in a non-expansion state (ST-04 §3.1), expecting a large subsidy, and found they qualified for nothing — too poor for premium tax credits, ineligible for Medicaid. The fix is to raise MAGI over 100% FPL with a Roth conversion. They had the right technique and no target.

💡 Pro-Tips

1. Rank by MAGI per dollar spent, not by tax rate — the ranking changes and the ACA cliff is why.
2. Know your Roth contribution basis to the dollar. It is your emergency fund, your bridge, and your most flexible asset, and most people underestimate it.
3. Spend zero-MAGI sources in this order: cash, HSA reimbursements, taxable basis, seasoned conversions — and Roth contributions last, because they are the only one that is also an emergency fund.
4. If you have a governmental 457(b), spend it before anything with a penalty, and do not roll it away.
5. Track each conversion’s five-year clock separately ✅ — they are not one queue with one date.
6. Set a MAGI target with a floor and a ceiling, not a minimum (HC-03, HC-01).
7. Use Roth conversions as the MAGI dial when your natural income lands too low — it costs no cash flow and builds the ladder.
8. Check whether a non-governmental 457(b) is a meaningful share of your net worth. It is an unsecured claim on your former employer ✅, not a retirement account in the ordinary sense.
9. Set cost-basis method to specific ID today, in the account settings. The FIFO default can triple the MAGI of an identical sale (§3).
10. If you are buying marketplace coverage and unemployed, the health-premium exception is uncapped ✅ and lands exactly where your costs are — it is the one § 72(t) exception worth planning around.

⚠️ Common Pitfalls

1. Believing Roth money is locked until 59½. Contributions never are ✅.
2. Minimising MAGI into the Medicaid floor or the coverage gap (§4).
3. Rolling a governmental 457(b) into an IRA and forfeiting its no-penalty status ◻️.
4. Confusing the two five-year clocks — one for earnings, one per conversion.
5. Assuming a non-governmental 457(b) is as safe as a 401(k). It is exposed to employer creditors ✅.
6. Spending Roth basis first because it is the easiest, and having no flexible asset left.
7. Selling taxable lots at the default FIFO setting, realising the largest gain available and the most MAGI — a $24,000 difference on one $40,000 sale (§3).
8. Treating a securities-backed line or margin as income. It creates no MAGI because it is debt, and the call arrives when markets are down (ER-06).
9. Forgetting the HSA shoebox exists and taking taxable withdrawals instead (HC-04).
10. Treating § 72(t) exceptions as a plan rather than as emergency hatches.
11. Building the bridge in the year you retire rather than five years before (TR-01).

✅ Actionable Checklist

Inventory (do this once, then maintain it)

• Roth IRA contribution basis to the dollar — from Form 5498s or custodian records
• Every Roth conversion with its year, so each five-year clock is dated
• Taxable lots with basis and holding period; set cost-basis method to specific ID (TX-01)
• HSA receipts accumulated but not yet reimbursed (HC-04)
• Whether any 457(b) is governmental or non-governmental — ask, do not assume
• Whether ER-04 Rule of 55 is available from your last employer’s plan

Each year

• Set a MAGI target — floor above 138% (or 100%, or 200% in a BHP state ✅), ceiling under your CSR or cliff line (HC-03, HC-01)
• Fund spending from the ranked stack in §2 until the target is met
• Use a Roth conversion to top up MAGI if you land under the floor
• Re-check that no rollover is about to destroy a 457(b) or Rule of 55 exemption

Sources & further reading (verified August 2026)

• IRC § 408A(d)(4) and IRS Pub. 590-B — Roth IRA distribution ordering rules: contributions, then conversions FIFO, then earnings; all Roth IRAs treated as one; contributions available tax- and penalty-free at any age
• IRC § 72(t)(2)(A) and the IRS “Retirement topics — Exceptions to tax on early distributions” table — the statutory exceptions, including those added or modified by SECURE 2.0
• IRC § 457(b) and IRS guidance on governmental versus tax-exempt 457(b) plans — the absence of the 10% additional tax on governmental plan distributions after separation, and the general-asset/creditor exposure of non-governmental plans
• FN-02 for the FPL, CSR, and cliff figures used in §3–§4 · ER-02 for the ladder that produces source 5 · ER-04 for the employer-plan route · ER-03 for the fallback when none of this reaches far enough · HC-03 for the floor this page can accidentally fall through