The History of the Safe Withdrawal Rate — The Trinity Study

The Trinity Study is a paper published in the February 1998 issue of the AAII Journal by three Trinity University professors, Philip Cooley, Carl Hubbard, and Daniel Walz.

Using actual U.S. stock and bond return data from 1926 through 1995, they calculated the probability that a retiree could withdraw a fixed percentage of their assets each year without running out of money over 30 years. They tested a range of withdrawal rates from 3% to 12%, combined with various asset allocations from 100% stocks to 100% bonds.

In this study, the definition of "success" is simple. If assets are greater than zero at the end of the 30 years (that is, money is left over), it counts as a success.

One caveat is that this definition does not mean "plenty was left over." Even if only $1 remains, it counts as a success. So even when the success probability is high, some cases are "barely made it" outcomes in which assets shrank dramatically. You have to keep this limitation in mind when looking at the success-probability number.

Looking at the key results, under the condition of adjusting the annual withdrawal amount for inflation each year, the 30-year success probability of a 4% withdrawal rose as the equity allocation increased.

As rough figures, a 50/50 stock-bond portfolio showed about a 95% success probability, and a 100% stock portfolio about 98%. The researchers concluded that "if you hold at least 75% stocks, an inflation-adjusted withdrawal of 4-5% is feasible."

What matters, though, is that this probability is a backtest against past U.S. data. Depending on future returns—especially whether a large crash occurs early in retirement—actual results can differ greatly.

Looking only at the 95% success probability is reassuring, but behind it lies a drawdown you must endure. The higher the equity allocation, the higher the success probability—but you also have to accept crashes of -30% or -50% during retirement.

In particular, the 4% rule assumes you "keep raising the withdrawal amount in line with inflation." If a bear market hits right after you retire, your assets shrink while your withdrawals hold steady, so your principal is eaten away quickly. This is called sequence-of-returns risk, and even for the same average return, whether the crash comes early or late determines the outcome. The success probability is just a number that averages this risk away; it does not eliminate the maximum drawdown and loss duration an individual will actually experience.