The 4% Rule Through Every 30-Year Retirement Since 1928: Four Failures, All Starting 1965 to 1969

The 4% rule says you can spend 4% of your portfolio in the first year of retirement, raise that amount with inflation every year after, and the money should last 30 years. It is quoted constantly and checked rarely. This guide runs it through every 30-year retirement that could have started between 1928 and 1996, using the annual returns of the S&P 500 and the 10-year Treasury bond and the actual change in prices, and looks at the few that failed. It uses the same engine as the retirement withdrawal calculator, so you can run your own numbers afterwards.

Three retirements that failed and four that did not

Balances are in the dollars of the day each retirement started, so they can be compared. “Ran short” means the year in which the withdrawal could no longer be paid in full.

The 1966 retiree was down to 49% of the starting buying power after ten years and 18% after twenty: the damage is done in the first decade, and the end is only the arithmetic of it. The 1929 retiree, who started on the eve of the Depression, ended with $524,000.

Why 1966 failed and 1929 did not

Because the danger is not a falling market alone but falling real value while you withdraw. Over the first ten years from 1929 prices fell, to 82% of their starting level, so each year's inflation-adjusted withdrawal was a smaller number of dollars; stocks fell hard (a dollar of them had lost more than half its buying power by the end of 1932, 0.46) but had recovered to 1.03 by the end of the tenth year. Over the first ten years from 1966 prices rose to 175% of their starting level and a dollar of stocks, dividends included, was worth 0.79: little real growth to draw on, with the withdrawals rising every year. Bonds were no refuge: 10-year Treasuries bought at the start of 1966 and rolled each year lost about 40% of their buying power by the end of 1981.

The same $40,000 produced very different retirements: a typical 1960s start had about $127,000 left after 30 years and a typical 1980s start $4,204,000, in each case in the dollars of the start year. A rule that survives both is working at the margin of the first and leaving a great deal of money in the second, which is why the amount you could have spent is better described as a floor than as a forecast.

What changes the odds

The share of start years in which 30 years of spending lasted, from $1,000,000, by stock share and first-year spending rate:

Too few stocks was the most expensive mistake: with a quarter in stocks the 4% rate lasted in 74% of start years against 94% at 60%. All stocks did no better than three-quarters (93% against 94%), because the bad start years punish a portfolio that has nothing steady to draw from.

Other income changes the question

Few people spend only from a portfolio for 30 years. If Social Security (or a pension) covers part of the spending from some year on, the portfolio carries a smaller load afterwards. Spending $50,000 a year with no other income is a 5% first-year rate and lasted in 72% of start years. With $20,000 a year of Social Security (in today's dollars, so it keeps up with prices) starting in year 8, the same $50,000 lasted in 100% of them. A bigger plan, $60,000 with $24,000 from year 8, lasted in 86%. The years before the income starts are where the risk sits, so the timing of the claim matters as much as the amount; the Social Security calculator shows what each claiming age pays.

What this guide leaves out

• One country. The United States did unusually well in these years; the same test on other countries' records is less kind. Any study built on a record that survived is built on a favourable sample.
• Yearly data and fixed spending. Returns are for calendar years, so a bad month inside a year is smoothed, and the spending never changes except with prices. Real retirees cut back in bad years, which makes the failures above less severe than they look.
• Taxes, fees and bonds. The figures are before tax and fees. The bond series is a 10-year Treasury rolled once a year, not a fund; a corporate or short-term mix would give different results.
• Overlap. Windows that start a year apart share 29 of their 30 years. The 69 results are one history seen 69 times, not 69 draws.

Not financial advice. These are descriptive results for past US returns; they do not predict what your own retirement will do.