The 4% Rule Isn't a Law of Physics. Here's the Math Behind It and Where It Breaks
7 min read
Loading...
7 min read
You've probably seen it written somewhere: retire, withdraw 4% of your portfolio a year, and your money should last. That single number gets repeated so often it starts to sound like a fixed law, the way gravity is 9.8 m/s². It isn't. The 4% rule is the worst-case result from one specific stretch of historical data, and even the researcher who calculated it now says the real number might be closer to 4.7%. Here's where the 4% rule actually comes from, the math behind it, and why applying it directly to an IHSG-heavy, rupiah-denominated portfolio needs a second look.
The rule traces back to financial planner William Bengen's 1994 paper, "Determining Withdrawal Rates Using Historical Data," and was popularised four years later by Philip Cooley, Carl Hubbard and Daniel Walz, three professors at Trinity University in Texas, in a paper now known simply as the Trinity Study.
What they actually did: take every rolling 30-year period in US market history, and for each one, ask "if a retiree withdrew X% of their starting portfolio in year one, then the same rupiah (or dollar) amount adjusted for inflation every year after, would the money have lasted the full 30 years?" Testing withdrawal rates from 3% to 12%, across portfolios split various ways between US stocks and bonds, 4% was the highest starting rate that survived every single 30-year window in the dataset, including the worst ones: retiring right before the Great Depression, or into the high-inflation 1970s.
That is the whole rule. It is not a formula that outputs 4% from first principles. It is the answer to "what would have survived the worst year to retire, in this one country's market history."
The mechanic itself is simple, and it is not what most people assume. The 4% rule does not mean withdrawing 4% of your portfolio's current value every year. It means calculating 4% once, in year one, and then increasing that same rupiah amount by inflation every year after, regardless of what the portfolio does.
Say you start retirement with a Rp2,000,000,000 portfolio. Here is what three years actually look like, using Indonesia's own July 2026 headline inflation rate of 2.88% year-on-year as reported by Bank Indonesia:
| Year | Withdrawal amount | How it was calculated |
|---|---|---|
| 1 | Rp80,000,000 | 4% × Rp2,000,000,000 |
| 2 | Rp82,304,000 | Rp80,000,000 × 1.0288 |
| 3 | Rp84,674,355 | Rp82,304,000 × 1.0288 |
Notice what is missing from that table: the portfolio's actual year-to-year performance. That is the point of the rule. It deliberately ignores whether the market was up or down and just keeps paying out the same real (inflation-adjusted) amount, on the assumption that a diversified, mostly-stock portfolio's growth will average out over the full 30 years even though any single year might not cooperate.
The 4% figure is specific to the exact 30-year windows, the exact asset mix (Bengen used a roughly 50-75% stock allocation against US Treasury bonds) and the exact historical returns and inflation path in the dataset it was tested on. Change any of those inputs and the number that survives every worst case changes too. That is not a flaw in the method, it is just what a backtest is: an answer conditioned on the data it ran against, not a constant of nature.
The clearest proof that it was never fixed is that its own author has since revised it. In his August 2025 book A Richer Retirement, Bengen raised his own recommended starting rate to 4.7%, after adding more asset classes to the model and factoring in market valuation and expected inflation at the time of retirement rather than using one flat number for every starting year. Morningstar's own 2026 retirement-income research goes the other way for its base case: a 3.9% starting rate for someone retiring in 2026 with a 30% to 50% equity allocation, up slightly from 3.7% the year before, but still below Bengen's figure because Morningstar's model assumes a more conservative, bond-heavy mix. Two well-resourced research teams, looking at broadly similar questions in the same year, land on 3.9% and 4.7%. If this were a law of physics, it would not move depending on who is doing the calculation.
The original backtest ran on US large-cap stocks and US government bonds. IHSG is a different market with a different risk profile entirely. Over 1984 to 2021, Indonesia's stock market averaged a 16.13% annual return, well above the long-run US equity average, but that headline number hides much rougher swings along the way: a 30.91% loss in 1998, during the Asian Financial Crisis, against a 187.43% gain in 1989. A rule calibrated to smooth out US-style worst cases was never tested against swings of that size, and no equivalent "Trinity Study for the IHSG," running the same 30-year rolling analysis on Indonesian market history, appears to have been published. Anyone applying 4% to an IHSG-heavy portfolio is borrowing a number that was never actually stress-tested against Indonesia's own worst-case years.
The 4% rule already has to account for the fact that when the bad years land matters more than their average size, a separate mechanic called sequence of returns risk. Two portfolios earning the exact same average return can end up tens of millions of rupiah apart purely because of which year the crash happened to fall in. The Trinity Study's worst-case testing implicitly covers this for its own dataset, but a rule of thumb copied out of context, without rerunning that worst-case check against a different market's own history, quietly drops this protection.
None of this means 4% is a bad number to think about. It means it is a starting reference point borrowed from a market that behaves differently from Indonesia's, built on assumptions (a specific stock and bond mix, US inflation history, a fixed 30-year horizon with no spending flexibility) that may or may not hold for your own situation. Bengen's revised 4.7% and Morningstar's 3.9% for 2026 both show that even inside a single, better-documented market, the "safe" number keeps being revised as researchers add more data, more asset classes and more realistic assumptions about how retirees actually spend.
For a rupiah portfolio, the honest version of the rule is closer to: pick a starting rate on the conservative side, given that IHSG's own worst-case years are less thoroughly studied than the US market's, and check that rate against your real holdings and volatility on a regular basis rather than setting it once and never looking again.
Treat 4% as a conversation starter, not a guarantee. Check your own portfolio's actual composition, drawdown history and volatility on NetWort's Portfolio Health view before picking any starting withdrawal rate, and see today's BI Rate and inflation reading on the Macro Context page, since both directly affect how far a fixed rupiah withdrawal will really stretch this year.