It should be clear by this point that the only answer to the “right” number for sustainable withdrawal rates is both vague and unhelpful: It depends. But we can make some educated guesses to test a plausible scenario. There’s no question that lower bond yields don’t bode well for future fixed-income returns. At the same time, high equity market valuations suggest the possibility of lower stock-index returns going forward. To account for both of these scenarios, we used real (inflation-adjusted) returns on stocks and bonds as a starting point but reduced the arithmetic averages for both asset classes by 200 basis points. That leads to average real return assumptions of about 7% for stocks and 0.4% for bonds.

To get a better sense of how portfolio results might play out with different withdrawal levels, my colleague Maciej Kowara used the real return assumptions above as a baseline (keeping volatility and correlation assumptions in line with the historical averages) and tested various initial withdrawal rates over a 30-year time horizon. He then ran a Monte Carlo analysis to randomly generate 10,000 potential return paths for each spending rate to generate a range of scenarios and estimate the probabilities of different outcomes.

As shown in Exhibit 1, these assumptions lead to a slightly lower sustainable withdrawal rate for a portfolio combining 50% stocks and 50% bonds. An investor looking for a 90% probability of success wouldn’t quite get there with the traditional 4% initial withdrawal rate, but ratcheting starting withdrawals down to 3.5% improved the odds. The table also shows the average length of the shortfall for portfolios that ran out of money before the end of the 30-year period.