ERN's Safe Withdrawal Rate series: a reader's guide

If this book could send you to one outside source on everything it covers, it would be this one: the Safe Withdrawal Rate Series on the blog Early Retirement Now (ERN). Karsten Jeske writes it, a PhD economist who worked at the Atlanta Fed and in quantitative asset management, and who retired early in 2018. More than sixty parts since 2016, every one of them built on reproducible simulations over monthly US data going back to 1871. Article after article, the series took apart the slogans of first-generation FIRE and laid down most of what is now said seriously about long retirements.

This page is a reader's guide: the major results, where to find them, and what to know about the author's leanings so you can read him well. Many articles in this book argue with the series; this one hands you the map.

The author and the method§

ERN's frame holds across the whole series. Monthly US data since 1871 (S&P composite stocks, 10-year government bonds, from Shiller). Simulated retirements starting in every possible month, over horizons of 30 to 60 years. And a close look at a criterion that is not just ruin, but ending wealth and the path too: when the failure hits, and what the retiree lived through on the way there. The tools are public. The Google Sheets toolbox of Part 28 lets you redo every calculation at home.

Two leanings are worth knowing before you start. The first sits in the data, which is American. The series inherits the optimism of that sample (Bengen, the Trinity study, and the birth of the safe withdrawal rate, Beyond the United States: Anarkulova, Cederburg and the world sample). ERN knows it, says so, and holds that 150 years of monthly data from a big market, 1929 and 1966 included, already discipline the conclusions hard. The second is a posture. ERN writes for the early retiree who refuses to depend on luck. Hence a robustness bar, often "survive the worst vintage", tougher than the 90 to 95% success rates practitioners work with (Modern guardrails (Morningstar): the state of the art, Failure probability: reading it, choosing it, and not letting it run you).

The major results, part by part§

The bedrock: 4% was not built for you (Parts 1 to 3, 26). This is the founding result. The 4% rule holds over 30 years. A horizon of 50 to 60 years calls for 3.25 to 3.5% if the withdrawal is rigid, meaning no flexibility at all on the amount, and the safe rate depends heavily on starting valuations (Part 3). Above a CAPE of 20, the historical SAFEMAX drops at every horizon. Part 26 ("Ten Things the Makers of the 4% Rule Don't Want You to Know") is the best popular summary (The 4% rule in ten minutes, Valuations, the CAPE, and what they say about your withdrawal rate).

Sequence explains everything (Parts 14 and 15). The series shows, with the numbers in hand, that the average return over 30 years matters less than the return of the first 5 to 10 years. That is the heart of Sequence of returns risk: the retiree's real enemy. Part 53 adds the corollary: the saver and the retiree face opposite exposures to sequence.

Flexibility is oversold (Parts 9 to 11, 23 to 25, 58). This is the most uncomfortable block of the series, and the most useful. The Guyton-Klinger rules (Parts 9 and 10) post flattering "success" rates. In the bad vintages, those rates are paid for with decades of spending cut by 30 to 45%. Ruin is simply swapped for a long stretch of poverty that no table of success rates shows (Guyton-Klinger: the original guardrails, their power and their limits). Parts 23 to 25 and 58 generalize the finding: any realistic flexibility, bounded and livable, buys a few tenths of a point of withdrawal rate, not the advertised magic (Flexibility: myth and reality (what it can really absorb)). That result is what pushed the whole field to show the standard of living delivered, and not just the survival of the portfolio (Using the FIRE simulator).

The CAPE as a rule, not as a fear (Parts 18 and 54). This is the series' constructive proposal: withdrawal rules built on the CAPE, where the rate, initial and current alike, adjusts to valuations, formalized in Part 54. It is the direct ancestor of the CAPE anchor in the FIRE simulator (CAPE-based rules: tying the withdrawal to valuations (ERN)).

Buckets and cash do not do what people think (Parts 12, 48, 55). This one cuts against the bucket dogma. A cash buffer drawn down and then mechanically refilled barely lowers the risk of ruin, because cash costs in return what it saves in sequence. And the popular bucket strategies are often market timing in disguise, with no clear rule (The cash buffer: size, cost, and what it is really for, The bucket strategy: the promise, and the case against it, Drawing on a buffer and refilling it: the rules that work). The nuance matters. Simulation lands in the same ballpark, with the buffer trade-off usually flat (The cash buffer: size, cost, and what it is really for). None of which takes anything away from what the buffer is worth psychologically, or for governance (The psychology of spending: why spending it is the hard part).

Glidepaths work (Parts 19, 20, 43). This is the good news of the series. Starting at 60% stocks and climbing back toward 100% over ten years or so buys 0.1 to 0.3 points of withdrawal rate in the worst vintages, at a modest cost in the good ones. The protection lands squarely on the fragile window (Glide paths: the bond tent, rising equity, and the fragile window, Accumulation, transition, withdrawal: the three lives of a FIRE plan).

Inflation, from both ends (Parts 41 and 51). Part 41 asks whether low inflation on the day you leave allows a higher rate. Part 51 answers that the level on that day predicts almost nothing, because it is the inflation of the years that follow that shaves the sustainable rate down, by roughly 0.2 points for every point of average inflation over the first decade (Inflation and the withdrawal rate: the exact link).

Annuities, bonds, gold, real estate, leverage (Parts 29 to 31, 34 to 36, 40, 49, 52, 56, 59). These are the asset reviews. The yield illusion of dividend portfolios, since living off your dividends is no safer (Parts 29 to 31 and 40, The ten mistakes that wreck a FIRE plan). Gold as a partial sequence hedge (Part 34, Gold in a retirement portfolio). Real estate (Part 36, Real estate in a FIRE plan (your home, and rentals)). Annuities and Social Security (Parts 56 and 59, Annuities and safety first: buying a floor). Leverage, sophisticated and in small doses (Parts 49 and 52, Leverage, margin and borrowing against your portfolio (advanced)).

The promises of a higher rate, one at a time (Parts 62 to 65). The latest parts put four recipes to the test, each one said to raise the safe rate. First small-cap value, a negative verdict (Part 62, 2025, Factors (Fama-French, value, momentum) in the withdrawal phase). Then tactical allocation by momentum, the only one that wins Jeske over, and it is a rule of his own making. Run on half the portfolio, it lifts the safe rate of the worst vintage to 4.24% over 30 years, against 3.84% for a 75/25, though he advises against most managed futures ETFs (Part 63, 2025, Managed futures and trend following: the diversification that works in a crisis). Then the risk parity portfolios, All Weather and the Golden Butterfly first among them, which he judges inferior to a plain 75/25 (Part 64, 2026, All-weather portfolios: Browne, All Weather, Golden Butterfly, Dragon). Last, Bengen's new five-style portfolio and its 4.7%, a gain he ties to the pre-1981 small-cap premium (Part 65, 2026, Factors (Fama-French, value, momentum) in the withdrawal phase). Jeske draws one lesson from all four: a building block that convinces you gets added to a stock-heavy portfolio, and it does not justify a recipe that waters the stocks down. Read these four parts through the last filter described below.

The side topics that change a life (Parts 21, 22, 27, 37, 42, 47, 60). The mortgage in retirement, where keeping the loan is a levered bet on sequence (Part 21). "One more year", with numbers attached (Part 42, The one-more-year syndrome). What deserves a fine calculation and what you can eyeball, from the rhythm of withdrawals to spending drift (Part 47, When to worry, and when to let it ride). Getting through a bear market (Part 37, Riding out a bear market in retirement: the playbook). And the critique of Die With Zero (Part 60, Real spending in retirement (the retirement smile, Die With Zero)).

Reading ERN with the right filters§

Four filters, and how much of each you need depends on where you live.

The geographic filter. If you are American, this one barely applies. The data, the tax chapters, Social Security, the 401(k) and IRA material all land in your own system, and you can read them straight (US taxes in the withdrawal phase). One caveat survives even then: the numbers inherit the sample. A century and a half of US returns is the record of the luckiest large market there has been, and the safe rates that come out of it stay on the optimistic side (Beyond the United States: Anarkulova, Cederburg and the world sample). Read from anywhere else, the filter does real work. The mechanisms carry over untouched, whether it is sequence, the CAPE, glidepaths or flexibility. The numbers need a check against your own currency and home market, and the tax and pension chapters have to be swapped for yours.

The posture filter. The bar "survive the worst vintage in the record" is a deliberate stance of caution, tougher than the 90 to 95% success practitioners settle for. If you have solid safety nets (a pension, employability, real flexibility), ERN's recommendations are a prudent bound for you, not a survival minimum (Failure probability: reading it, choosing it, and not letting it run you, How much you need).

The perfectionism filter. The series proves that you can refine forever, down to the tenth of a point of SWR and the allocation within 5 points. Keep one reminder in mind as you read. The first three levers of a real plan are still audited spending, the pension counted, and a written rule (The ten mistakes that wreck a FIRE plan). Refinement comes after those.

The assumption filter. The recent parts rule more bluntly than the early ones, in a tone that likes a fight. Their conclusions still rest on choices you can argue with. Recent funds are rebuilt with a factor model fitted on about twenty years of ETF data. Future premia are set to zero, which is an assumption and not a finding. The momentum rule of Part 63 is the author's own, judged on the history it was designed on. And gold is replayed before 1971, when its price was officially fixed. These are serious second opinions, to be read as such, not verdicts.

The essentials§

Going further§