[TX-07] Guardrails and Dynamic Spending Rules

The rules that let you start higher than 4% — and the cut they will ask for in the year you least want to hear it.

Pillar: Tax Optimization & Decumulation · Applies to: Anyone spending from a portfolio who wants a written rule instead of an annual argument, at any age Last verified: August 2026 · Refresh cadence: Evergreen — these are published rules, not annual figures Related: ER-01 Safe Withdrawal Rates · ER-06 Sequence of Returns Risk · TX-03 Withdrawal Sequencing · TX-06 Senior Deduction · HC-01 ACA Bridge · HC-06 IRMAA · SS-01 Claiming Age Math

Not advice. Every rule here will, in a bad decade, instruct you to cut real spending substantially — by 42% at the trough in §2’s simulation, and by more than half in the worst historical cohorts. That is the mechanism working, not failing. Do not adopt a rule whose worst case you have not looked at and cannot live with.

---

TL;DR

• A spending rule is what lets you start above 4% (ER-01). Guyton and Klinger’s own claim is 5.2%–5.6% with at least 65% in stocks ✅ — bought entirely by agreeing in advance to cut when the portfolio falls.
• On the identical market ER-06 used, the fixed 4% plan ends at $382,840 and Guyton-Klinger ends at $3,150,668 — but delivers $890,292 of real spending against the fixed plan’s $1,200,000. The rule did not create money. It moved consumption from the retiree to the estate.
• Rank on spending delivered and worst year lived, not on ending balance. Every comparison leads with the column that makes flexible rules look free.
• Guyton-Klinger and a deliberately crude rule reach the identical worst year — $23,317, a 42% real cut. The published rule’s sophistication buys a smoother path to the same floor, plus 17% more total spending; it does not buy a shallower floor.
• Kitces’s critique is the honest counterweight: in historical cohorts the same rule cut real spending 28% (2007), 36% (dot-com), 45% (1936) and 54% (1965) ✅. You are trading a small risk of ruin for a moderate risk of large, sustained cuts.

---

1. What a rule actually buys

A fixed real withdrawal plan has one virtue — predictable spending — and one fatal property: it takes no information from the market. ER-06 showed what that costs when the order of returns goes badly.

A decision rule reverses the trade. You accept variable spending, and in exchange you may start higher, because the plan can respond before it fails. That is the entire economic content of every rule on this page, and it is why ER-01 §5 insists the variant matters more than the starting rate.

The rules divide into three families:

Family	Mechanism	Character
Guardrails	Do nothing until the withdrawal rate strays far from plan, then make a large discrete adjustment	Rare, big moves
Bands / dynamic spending	Recompute from the current balance every year, but clamp the year-on-year change	Frequent, small moves
Ratchets	Never cut; raise only after sustained good performance	One-directional

2. Guyton-Klinger, implemented and priced

The 2006 rules (Journal of Financial Planning, March 2006) are four, and the summaries usually mangle two of them:

Rule	What it does
Withdrawal rule	Take the CPI increase each year, capped at 6% ✅ — but skip it entirely in a year following a negative return if the current withdrawal rate already exceeds the initial rate
Capital preservation	If the current withdrawal rate rises 20% above the initial rate, cut dollar withdrawals 10% ✅
Prosperity	If it falls 20% below the initial rate, raise withdrawals 10% ✅
Portfolio management	Fund withdrawals from the asset class that outperformed; rebalance the remainder

The detail that matters and is usually lost: both guardrails compare against the initial withdrawal rate, not against last year’s. That is why the rule can cut in several consecutive years — each cut is measured against a fixed reference, so a portfolio that keeps falling keeps triggering. Capital preservation is suspended in the final 15 years ✅, on the reasoning that a plan near its end no longer needs preserving.

Run on the exact market ER-06 used — $1,000,000, $40,000 first-year spending, 3% inflation, bad decade first, the same thirty returns imported rather than retyped:

Rule	Ending balance	Real spending delivered	Worst real year
Fixed 4% real	$382,840	$1,200,000	$40,000 (0% cut)
Guyton-Klinger	$3,150,668	$890,292	$23,317 (42% cut)
Ceiling/floor bands	$2,284,930	$941,429	$24,726 (38% cut)
ER-06’s crude persistent cut	$3,775,458	$759,126	$23,317 (42% cut)

Read the middle column before the first one. Guyton-Klinger turned a near-failure into a $3.1M estate — and handed the retiree $310,000 less real spending over thirty years than the fixed plan did. The rule did not create money; it moved consumption from the household to its heirs. For a retiree whose goal is a bequest, that is a feature. For one whose goal was to spend their savings, it is the opposite of what they thought they were buying.

And note where the two cutting rules land: the identical worst year, $23,317, to the dollar. Both apply five 10% cuts and freeze inflation in the same years, so the cumulative trough is the same. What the published rule buys is the path, not the depth — it descends more gradually and delivers 17% more total spending on the way down. Sophistication in a spending rule smooths the ride; it does not change the destination a bad decade sends you to.

3. The critique, which belongs on the same page as the rule

Kitces’s analysis of the same rules in historical cohorts found the deepest real spending cuts to be 28% in the 2007 financial crisis, 36% after the dot-com peak, 45% for the 1936 cohort, and 54% in the 1965 stagflation cohort ✅. His summary is the fairest one-line statement of the trade: you replace a small risk of running out of money with a moderate risk of large spending cuts across many years.

The structural criticism is over-correction. Guardrails trigger on the withdrawal rate, which is a crude proxy for whether the plan is actually in trouble — it ignores how many years remain, what other income is arriving, and how the portfolio is invested. A rate-based trigger fires the same way at year 3 and year 25, though the situations are not comparable.

Risk-based guardrails trigger on estimated probability of success instead, and in the same cohorts produced cuts of 3%, 0%, 8% and 32% respectively ✅ — materially gentler for the same portfolio survival. They require a planning tool to compute, which is their real cost: you cannot run one on the back of an envelope, and that is a genuine argument for the cruder rule if the alternative is no rule at all.

4. Choosing one

If your budget is…	Use	Because
Mostly discretionary	Guardrails	Rare large moves are tolerable and buy the highest starting rate
Mostly fixed costs	Bands / dynamic spending	Small frequent moves; a 10% cut may be impossible for you
Already covered by guaranteed income	A ratchet, or no rule	Your portfolio is funding upside, not essentials (SS-01)
Managed with a planner or a tool	Risk-based guardrails	Gentler cuts for the same survival, at the cost of needing the tool

Three questions decide it, and none is about the rule:

1. What fraction of your spending can actually fall? If fixed costs are 70% of the budget, a 42% total cut is not available to you and a rule that assumes it is will simply be abandoned mid-decline.
2. How much of your spending is portfolio-funded at all? Guaranteed income shrinks the problem before any rule applies (SS-01, ER-01 §7).
3. Do you want to spend the money or leave it? §2’s table answers this question honestly and most comparisons do not.

5. Where the tax layer changes the rule

A spending rule sets how much you take. This wiki’s other pages set from where and what it does to your income, and the interactions are real:

• A cut in withdrawals is not a cut in MAGI if the withdrawals come from a taxable account with a low basis — and MAGI is what decides your ACA subsidy (HC-01) and your IRMAA tier (HC-06). Spending less does not automatically mean reporting less.
• The reverse also holds: a prosperity-rule raise can push you through a threshold. A 10% spending increase funded from a traditional IRA is a 10% increase in ordinary income, and near the 400% FPL cliff that is the most expensive raise available (HC-01 §4).
• Guardrail cuts arrive in bad markets, which is exactly when Roth conversions are cheapest — depressed balances convert more shares per dollar of tax. A year the spending rule says “cut” is often a year the conversion plan says “convert” (TX-02).
• Fund the cut from the right account. TX-03 owns the ordering; the point here is that a rule which lowers withdrawals without regard to source can leave a 0% capital-gains bracket unused (TX-01) or a zero bracket unfilled (TX-06).

The rule and the tax plan are separate systems, and only one of them is automatic.

Three Scenarios

A. Retired at 62, 70% of spending is fixed. Mortgage, insurance, property tax. A 10% guardrail cut is not something she can execute — the money is committed. Bands are the right family for her, and the honest conclusion is that she should have started nearer 3.5% than 5.5%, because the flexibility that justifies the higher rate is flexibility she does not own (ER-01 §4).

B. Retired at 66 with Social Security covering 60% of spending. Only 40% of his budget is exposed to the market at all, so a 42% cut in portfolio spending is a 17% cut in total spending. He can run an aggressive rule precisely because his floor is already built (SS-01) — the rule’s severity is buffered by income the rule does not touch.

C. FIRE at 44, all portfolio-funded, 50-year horizon. The rules were calibrated on 30-year retirements and the guardrails do not know his horizon (ER-01 §3). His real protection is not the rule — it is that he can earn again, which no rule models and which dominates every adjustment in §2. He should still write one down, because the point of a written rule is to survive the month he does not want to follow it.

💡 Pro-Tips

1. Write the rule down before you need it, with the trigger, the size, and how long it persists. A rule invented during a decline is a rationalisation.
2. Look at the worst case before adopting the rule, not the average case. §2’s worst year is the number that decides whether you will actually follow it.
3. Rank candidate rules on spending delivered and worst year lived — never on ending balance alone.
4. Match the family to your budget: guardrails for discretionary budgets, bands for fixed ones.
5. Compute your guardrails as dollar amounts, today, and put them in the plan. “Cut 10% if the rate rises 20%” is not something anyone evaluates correctly under stress.
6. Remember the reference is the initial rate, not last year’s — that is what allows consecutive cuts, and it surprises people in year three.
7. Check every raise against the ACA and IRMAA thresholds before taking it (HC-01, HC-06).
8. Treat a guardrail cut year as a conversion opportunity — depressed balances make Roth conversions cheaper in exactly those years (TX-02).

⚠️ Common Pitfalls

1. Adopting a rule for its higher starting rate without pricing its worst cut. The 5.2%–5.6% and the 42% are the same decision ✅.
2. Judging rules by ending balance, which rewards whichever rule spent least (§2).
3. Believing a sophisticated rule cuts less deeply. Here it reached the identical trough as a crude one; it only got there more gently.
4. Comparing the guardrail against last year’s rate instead of the initial rate — this understates how often the rule fires.
5. Assuming lower spending means lower MAGI. Source decides that, not amount (TX-03, HC-01).
6. Taking a prosperity-rule raise through a subsidy cliff and paying five figures for a 10% spending increase.
7. Running a rule calibrated for 30 years over a 50-year retirement (ER-01 §3).
8. Ignoring that capital preservation switches off in the final 15 years ✅ — the rule deliberately stops protecting late, which is not a bug but does change the late-retirement risk.
9. Adopting a rule you cannot execute because your budget is mostly fixed. It will be abandoned in the first bad year, which is worse than never having had one.
10. Forgetting non-portfolio income when sizing the cut. Scenario B’s 42% portfolio cut is a 17% lifestyle cut.

✅ Actionable Checklist

Choosing the rule (once)

• Compute what share of spending is genuinely discretionary
• Subtract guaranteed income; the rule applies only to the remainder (SS-01)
• Pick a family from §4 and write down the trigger, size, and persistence
• Run your rule’s worst case and confirm you could live on it

Converting it to dollars (once, then annually)

• Record the initial withdrawal rate — every guardrail references it forever
• Compute the upper and lower guardrail as dollar portfolio values, not percentages
• Diarise the annual review date and keep it in bad years especially

Each year

• Recompute the withdrawal rate against the current balance
• Apply the rule as written; record what it said even if you override it
• Check any increase against the ACA cliff and IRMAA tiers (HC-01, HC-06)
• If the rule cut this year, check whether a Roth conversion is unusually cheap (TX-02)
• Fund the resulting withdrawal in the right order (TX-03)

State notes (→ ST-01, ST-04): The rule is federal-agnostic; its tax consequences are not. A prosperity-rule raise funded from a traditional IRA is fully taxable in the 41 states with an income tax and free in the nine without ✅, so the same rule has a materially different after-tax cost by residence (ST-02). States with retirement-income exclusions change the calculus again — where pension or IRA income is partly excluded up to a threshold ◻️, a raise that crosses that threshold carries a state-level marginal cost the federal brackets do not show (ST-04 §1). Washington’s capital-gains excise tax ✅ means a rule funded by selling appreciated taxable holdings can trigger a state charge that a rule funded from an IRA would not — an unusual case where the source of the withdrawal, not its size, creates the state liability (TX-01, ST-01). Confirm your own state’s treatment before assuming a cut or a raise costs what the federal arithmetic says.

Sources & further reading (verified August 2026)

• Guyton, J. and Klinger, W., “Decision Rules and Maximum Initial Withdrawal Rates” (Journal of Financial Planning, March 2006) — the four decision rules, the ±20% guardrail triggers on the initial withdrawal rate, the 10% adjustments, the 6% cap on inflation increases, the suspension of capital preservation in the final 15 years, and the claimed 5.2%–5.6% maximum initial withdrawal rate at 65%+ equity.
• Kitces, M., “Why Guyton-Klinger Guardrails Are Too Risky For Retirees” — the historical worst-case real spending cuts of 28% (2007), 36% (dot-com), 45% (1936) and 54% (1965), the over-correction critique of rate-based triggers, and the risk-based alternative producing 3%, 0%, 8% and 32% in the same cohorts.
• Kitces, M., “Ratcheting The Safe Withdrawal Rate” (2015) — the one-directional rule: raise spending 10% when the portfolio exceeds 150% of its initial nominal value, no more than once every three years.
• Vanguard, dynamic spending research — the ceiling/floor formulation, commonly a +5% ceiling and −2.5% floor applied to the prior year’s real spending, which trades frequent small adjustments for rare large ones.
• tools/tx07_worked_examples.py — all four rules implemented and compared. The return series is imported from er06_worked_examples.py, not copied, so ER-06 and this page provably run the same market and cannot drift apart.
• ER-01 owns the starting rate this page justifies raising · ER-06 owns the risk these rules respond to · TX-03 owns which account funds the withdrawal · HC-01 and HC-06 own the thresholds a raise can cross.