Withdrawal strategies: the map before the territory

You have a portfolio, a horizon and an income to fund. One decision now governs the whole withdrawal phase: what rule, exactly, will you draw by? How much in year one? And above all, how will that amount respond to markets, to inflation, and to the years going by?

A dozen or so named answers exist: Bengen, the fixed percentage, Guyton-Klinger, VPW, CAPE rules, Morningstar guardrails, ABW/TPAW, Vanguard's floor and ceiling, annuities. The literature has been comparing them for twenty-five years. The good news is that the zoo is smaller than it looks. All of it turns on a single three-way tradeoff. Once you have the map, each rule becomes a point you can place at a glance, with predictable strengths and predictable pathologies.

This article is that map. It lays out the impossible triangle of withdrawals, the two extreme rules that bound the space, the five families and the article devoted to each, the six criteria that let you score a rule honestly, and how to read the decumulation frontier, the chart that places every rule against every other. The ten articles that follow take each rule apart. This one keeps you from getting lost among them.

The two pure extremes, and the frontier between them§

Start with the two strategies that bound the whole space. Each one delivers two corners of the triangle perfectly, by abandoning the third.

The income-first extreme: the fixed inflation-adjusted amount (Bengen's rule, The fixed inflation-adjusted withdrawal (Bengen): the benchmark rule). You take $X in year one, then the same amount adjusted for inflation, whatever happens. Income is perfectly stable, by construction. The price is that ruin becomes possible: if markets disappoint for long enough, the portfolio runs out before you do (The 4% rule in ten minutes, Sequence of returns risk: the retiree's real enemy). All the risk piles onto one event, binary and far away.

The capital-first extreme: the fixed percentage (The fixed percentage of the portfolio: indestructible but uncomfortable). Each year you take Y% of the portfolio as it stands. Ruin becomes impossible, by construction: Y% of something is never all of it. In exchange, income tracks the market one for one. A portfolio down 35% buys a standard of living down 35%, possibly for years. All the risk spreads out over daily life.

Between those two poles lies a continuum, and it gives this whole part its key chart, the decumulation frontier. Every rule becomes a point on it, fixed by two coordinates and no more: its failure probability, and the swing in living standard it imposes. Bengen sits at the bottom right, maximum failure and zero variability. The fixed percentage sits at the top left, zero failure and maximum variability. The rules in between trace an arc. The picture makes twenty-five years of research visible at once. You do not remove the risk from a withdrawal plan, you choose its shape: a rare and brutal bankruptcy on one side, frequent but bounded adjustments on the other.

The rules that listen are not magic. They simply sit better on the arc, closer to the ideal corner where failure and variability are both low. They get there because they listen to information the extremes ignore: the current portfolio, valuations, the horizon that is left.

There is nothing mysterious about the chart, and you can draw it yourself. A withdrawal engine only ever runs one spending policy at a time, so you get the frontier by running the full plan once per candidate rule and writing down the pair it returns, failure and variability. The scatter you accumulate is your plan's frontier. A few tools plot it for you, but any simulator that lets you swap the withdrawal rule will give it to you in a handful of runs. Two precautions keep the result readable. Hold the inputs and the market model strictly fixed from run to run, or you end up comparing rules and worlds at the same time. And settle on one measure of variability once and for all: the standard deviation of annual real spending, the spread of the worst quartile, the count of years spent below the comfort floor, any of them works, but mixing them across rules makes the numbers meaningless.

The full taxonomy§

Here is the zoo, sorted into five families by the information each rule listens to. The table is the map, and each rule gets its own article.

Three things stand out.

The more information a rule listens to, the better it sits on the frontier, and the more governance it demands. The fixed rule demands nothing, and learns nothing. Guardrails demand that you take cuts decided in advance, while calm, at the worst emotional moment. Amortization demands an annual recalculation and an income that is officially variable. The gain in position on the frontier is real and measurable. The cost is behavioral, and it is real too (The psychology of spending: why spending it is the hard part).

Families 3 and 4 have won the recent research. The modern consensus (Morningstar, Kitces, ERN, the Bogleheads, the actuarial literature) has moved. The pure fixed rule teaches well but is dominated. The pure percentage is indestructible and unlivable. The rules that listen, well-bounded guardrails and smoothed amortization, buy the best trade between failure and the life you actually live. The literature prefers the ABW/TPAW family for its internal consistency: it is the only one that can neither run dry early nor die on a pile of gold (Amortization-based withdrawal (ABW/TPAW): the actuarial approach). Practitioners still prefer guardrails, because a client can read them (Modern guardrails (Morningstar): the state of the art).

Family 5 changes the game. Safety-first is not hunting for a better point on the frontier. It takes part of the spending out of the triangle altogether, the vital floor, now covered by an annuity or a pension. The portfolio then carries only the comfort layer, where variability is tolerable. Wade Pfau is the school's contemporary theorist, and his line sums it up: retirement is not a portfolio problem, it is an asset-liability matching problem. Two camps shape the whole professional debate, those who optimize the probabilistic frontier (probability-based) and those who guarantee the floor first (safety-first). The best practical answer is usually a hybrid, and most real plans already are one: Social Security covers the floor, the portfolio funds comfort (Pensions and other income in the plan, US health coverage and Social Security: the two ends of the gap).

Scoring a rule honestly: the six criteria§

Comparing rules takes more than a success rate. Judge on that alone and the fixed percentage always "wins" (zero failure!) while sweeping its pathology under the rug. Here is the scorecard, inherited from ERN (Part 11) and from practice (ERN's Safe Withdrawal Rate series: a reader's guide):

1. Failure (at a given horizon, under a given model): the classic criterion, necessary and never sufficient (Failure probability: reading it, choosing it, and not letting it run you).
2. The average standard of living delivered: how much the rule actually lets you spend, on average, over the life of the plan. Two rules with the same failure probability can differ by 15% in total consumption.
3. The distribution of living standards in the bad scenarios: the criterion that really separates the rules. What does the rule deliver in the worst quartile, how many years below the comfort floor, and how far below? Answering means leaving the aggregates behind and reading the spending delivered year by year, along the harshest paths. That reading is what unmasked Guyton-Klinger: superb success rates paid for by decades cut 30 to 45% in the 1966 vintage (Guyton-Klinger: the original guardrails, their power and their limits, Flexibility: myth and reality (what it can really absorb)).
4. The bequest (the distribution of ending wealth): some rules die rich every time (the cautious fixed rule), others spend everything (amortization over an exact horizon). Neither is "right", but you should know which one you signed up for (Real spending in retirement (the retirement smile, Die With Zero)).
5. Governance: can a human under stress run this rule, and can a surviving spouse? A seven-parameter rule recomputed every month is a promise that it will not be followed (FIRE as a couple, and as a family, The annual review: the retiree's checklist).
6. Robustness to bad assumptions: what becomes of the rule if μ was overstated by a point? The rules that listen correct themselves along the way, which is their deepest strength. The fixed rule absorbs the error in silence, all the way to the cliff.

The map in action: one plan under four rules§

Nothing beats a worked case. The plan: $1.5M, a comfort need of $54,000 a year (3.6%), a floor set at $42,000 (How much you need), a 45-year horizon, and a pension of $15,000 a year starting in year 17. Here are the four points on the frontier (indicative numbers from the central model, run your own):

• Bengen, $54,000 indexed: failure about 9%. Perfectly stable income, right up to the cliff. Comfortable median bequest. The reference point.
• Fixed percentage, 3.6%: failure 0%. But the worst quartile spends years below the $42,000 floor as soon as the first hostile regime arrives, and income swings by plus or minus 25% from one decade to the next. Unlivable for this household.
• Guardrails (a −10% cut when the current withdrawal rate passes 4.5%, floor at 78% of comfort): failure about 3%, income stable most of the time, and in the bad quartile two to four cuts, never below the stated floor. Average lifetime spending comes out 4% below Bengen's.
• ABW (amortized over the remaining horizon, central return, smoothed): failure structurally near zero, the highest average total consumption of the four (the rule dares to spend what the others hoard), income officially variable but bounded by the pension and by the smoothing, and a small bequest, accepted up front.

The lesson is not that ABW wins. It is that the answer depends on the household. This one has a high floor (78% of comfort) and wants to spare the spouse who will have to run the plan, so plain guardrails take it. A flexible household with no interest in leaving a bequest would have taken ABW. A household that flinches at any variation would have taken Bengen at 3.2%, with more capital. The full selection work, criterion by criterion and profile by profile, is Choosing your strategy: criteria, comparison, worked case, the closing article of this part.

How to read the rest of this part§

The articles follow the map. The extremes come first, because they bound everything: The fixed inflation-adjusted withdrawal (Bengen): the benchmark rule, then The fixed percentage of the portfolio: indestructible but uncomfortable. Guardrails come next, from the historical classic (Guyton-Klinger: the original guardrails, their power and their limits, worth reading for the mechanism and for its pathologies) to the modern versions (Floor and ceiling and the Vanguard rules: bounded flexibility for the Vanguard family, Modern guardrails (Morningstar): the state of the art for the practitioner state of the art). Then the actuarial family: VPW, the Bogleheads' variable percentage withdrawal (the bridge from proportional to actuarial), Amortization-based withdrawal (ABW/TPAW): the actuarial approach (the literature's favorite) and CAPE-based rules: tying the withdrawal to valuations (ERN) (valuations as information, Valuations, the CAPE, and what they say about your withdrawal rate). Then the change of game: Annuities and safety first: buying a floor (the guaranteed floor). Then the seven rules side by side, on retirements that actually happened, year by year (Seven ways to live on one portfolio), and the decision summary, Choosing your strategy: criteria, comparison, worked case. Each article gives the exact mechanics, the parameters the research recommends, the known pathologies, and how to run the rule in practice. All of them fit in a few lines of arithmetic, and a serious simulator will swap one for another without changing anything else in the plan. That is exactly what you need to place them on the frontier one at a time.

The essentials§

• The impossible triangle, stable income, safe capital, high income: every rule gives up one corner, and "the best rule" does not exist until you say which sacrifice you are choosing.
• The two extremes bound the space: Bengen (stable, ruin possible) and the fixed percentage (indestructible, unlivable). Every other rule lives on the frontier between them, where each comes down to a pair of numbers, failure and the variability of the standard of living.
• Five families, sorted by what they listen to: fixed, proportional, guardrails, actuarial, guaranteed floor. Recent research favors well-bounded guardrails and amortization, practitioners favor how readable guardrails are, and the guaranteed floor changes the game instead of moving along the frontier.
• Six criteria for scoring a rule: failure, average standard of living, the life lived in the worst quartile (the one that separates them), bequest, governance, robustness to bad assumptions. Any table showing only a success rate is hiding what matters.
• The same plan changes its best rule depending on the floor, the bequest you want and your tolerance for adjustments. The reasoned selection is in Choosing your strategy: criteria, comparison, worked case.

Going further§

• Early Retirement Now, Part 11 (the criteria for scoring dynamic rules): the founding scorecard (ERN's Safe Withdrawal Rate series: a reader's guide).
• Wade Pfau, Safety-First Retirement Planning (2019): the liability-matching school, against the probability-based one.
• Morningstar, The State of Retirement Income: the annual comparison of the rules, scored on the life actually lived (Modern guardrails (Morningstar): the state of the art).
• Bogleheads wiki, "Withdrawal methods": the community inventory, well maintained.
• In this book: Seven ways to live on one portfolio (the seven rules replayed on three real retirements), Choosing your strategy: criteria, comparison, worked case (the reasoned selection, criterion by criterion) and Under the hood: how the simulator computes this book (how a simulator runs these rules and draws the frontier).