A well-known set of financial studies seems to show that a retiree can withdraw a fixed dollar amount from his investment portfolio that is about 4% of her initial portfolio value and avoid financial ruin for at least 30 years 95% of the time. For example, a retiree with a million dollar stock and bond portfolio on the day he retires could spend $40,000 annually and have a 95% probability of avoiding ruin for at least 30 years.

The studies back-test the following strategy using actual annual stock market returns since the 1800’s: A retiree earns the market return on his portfolio each year and spends a fixed dollar amount each year that equals 4% of his initial (not current) portfolio value and—importantly—continues this rate of spending until the portfolio is depleted, even when pending financial ruin seems obvious. (A retiree who expects to live 10 more years, for example, but whose portfolio will only support two more years of spending at this rate will spend this same amount, regardless.)

The model is quite simple. It takes a portfolio value, calculates 4% (or similar) annual spending amount. It then applies an historical market return to calculate annual earnings and subtracts the dollar spending amount. It uses rolling 30-year market returns and continues this process until the portfolio value is depleted.

The studies conclude that a “typical” retiree can therefore spend 4% of initial portfolio value with a 95% chance of successfully funding at least 30 years of retirement. Isn’t this, though, a case of the unrepresentative sample fallacy, given that any rational retiree would reduce spending to avoid bankruptcy? We cannot, I assume, conclude that rational retiree spending behavior would display bankruptcy rates similar to this sample group.

The 5% bankruptcy rate predicted for this strategy is 12 times the actual national average for bankruptcy claims by retirement-aged households, by the way.

Is this model as flawed as it appears to me to be? Thanks.