Reading on withdrawal approaches, I kept coming across the often-quoted guideline for using age to set stock allocation, traditionally "100% - Age" and more recently sometimes "120 - Age". I was curious how these age-based guidelines performed using historical data, so I ran some tests. IMO the results show age-based guidelines do not perform well in retirement. (On one Boglehead threat, a poster stated the age-based guidelines are only for the accumulation phase, not retirement, but I didn't see that caveat mentioned elsewhere.)
The tests use Shiller's data (i.e. S&P index and long interest rate from 1871 to present). All tests assume a 30-year retirement starting at age 60 with an inflation-adjusted annual withdraw amount and annual rebalancing. For the "100-Age" case, the stock allocation starts at 40% then decreases each year to end at 10%. For "120-Age", the stock allocation starts at 60% and ends at 30%. The below results compare the two age-based approaches to fixed allocations (for which either the equity percentage or the withdraw rate was varied to match the age-based success rate for a "break-even" comparison.) The "Success Rate" indicates the percentage of rolling 30-year periods when the portfolio did not run out of money....the same approach as FIRECalc.
4.00% Withdraw for 30 Years and 100-Age for Equities --- 88.99% Success Rate
4.00% Withdraw for 30 Years and Fixed 28% Equities --- 88.99% Success Rate
4.22% Withdraw for 30 Years and Fixed 40% Equities --- 88.99% Success Rate
4.00% Withdraw for 30 Years and 120-Age for Equities --- 94.50% Success Rate
4.00% Withdraw for 30 Years and Fixed 56% Equities --- 94.50% Success Rate
4.12% Withdraw for 30 Years and Fixed 60% Equities --- 94.50% Success Rate
5% Withdraw for 30 Years and 100-Age for Equities --- 44.95% Success Rate
5% Withdraw for 30 Years and Fixed 32% Equities --- 44.95% Success Rate
5.34% Withdraw for 30 Years and Fixed 40% Equities --- 44.95% Success Rate
5.00% Withdraw for 30 Years and 120-Age for Equities --- 60.55% Success Rate
5.00% Withdraw for 30 Years and Fixed 50.8% Equities --- 60.55% Success Rate
5.20% Withdraw for 30 Years and Fixed 60% Equities --- 60.55% Success Rate
For a higher-level view, the below diagram shows the maximum successful withdraw rate for each year using the two age-based guidelines compared to fixed allocations.
View attachment aged-based withdraws.pdf
Some thoughts....
-- Volatility risk is always reduced by lowering the equity ratio (assuming high-quality bonds), but the real risk we are concerned with is running out of money (although volatility can certainly contribute to running out of money). These different measures for risk seem to get blurred together and mixed up.
-- The age-based guidelines resemble a stocks-first harvesting strategy, which has been shown elsewhere to not perform well, so in this sense the results are not surprising.
-- I would not read too much into the results (although I find them interesting), just that if we lock in a withdrawal plan at the start of retirement, with blind adherence for the life, then a fixed stock/bond ratio performs better than an age-based ratio. In real life we don't blindly adhere to a plan, but reevaluate and adapt as we grow older.
-- We often should reduce equities as we age to lower retirement risk, but that decision should come from a broad-based reevaluation as circumstances change....and then only using guidelines with clear evidence to back them up. For example, there is evidence immediate annuities can reduce risk as we age. Another solid approach could be restarting/recalculating a withdrawal schedule at a later age using a shorter duration to derive a different stock/bond mix.
-- I doubt many here use an age-based formula, but not sure.
Thoughts....?
The tests use Shiller's data (i.e. S&P index and long interest rate from 1871 to present). All tests assume a 30-year retirement starting at age 60 with an inflation-adjusted annual withdraw amount and annual rebalancing. For the "100-Age" case, the stock allocation starts at 40% then decreases each year to end at 10%. For "120-Age", the stock allocation starts at 60% and ends at 30%. The below results compare the two age-based approaches to fixed allocations (for which either the equity percentage or the withdraw rate was varied to match the age-based success rate for a "break-even" comparison.) The "Success Rate" indicates the percentage of rolling 30-year periods when the portfolio did not run out of money....the same approach as FIRECalc.
4.00% Withdraw for 30 Years and 100-Age for Equities --- 88.99% Success Rate
4.00% Withdraw for 30 Years and Fixed 28% Equities --- 88.99% Success Rate
4.22% Withdraw for 30 Years and Fixed 40% Equities --- 88.99% Success Rate
4.00% Withdraw for 30 Years and 120-Age for Equities --- 94.50% Success Rate
4.00% Withdraw for 30 Years and Fixed 56% Equities --- 94.50% Success Rate
4.12% Withdraw for 30 Years and Fixed 60% Equities --- 94.50% Success Rate
5% Withdraw for 30 Years and 100-Age for Equities --- 44.95% Success Rate
5% Withdraw for 30 Years and Fixed 32% Equities --- 44.95% Success Rate
5.34% Withdraw for 30 Years and Fixed 40% Equities --- 44.95% Success Rate
5.00% Withdraw for 30 Years and 120-Age for Equities --- 60.55% Success Rate
5.00% Withdraw for 30 Years and Fixed 50.8% Equities --- 60.55% Success Rate
5.20% Withdraw for 30 Years and Fixed 60% Equities --- 60.55% Success Rate
For a higher-level view, the below diagram shows the maximum successful withdraw rate for each year using the two age-based guidelines compared to fixed allocations.
View attachment aged-based withdraws.pdf
Some thoughts....
-- Volatility risk is always reduced by lowering the equity ratio (assuming high-quality bonds), but the real risk we are concerned with is running out of money (although volatility can certainly contribute to running out of money). These different measures for risk seem to get blurred together and mixed up.
-- The age-based guidelines resemble a stocks-first harvesting strategy, which has been shown elsewhere to not perform well, so in this sense the results are not surprising.
-- I would not read too much into the results (although I find them interesting), just that if we lock in a withdrawal plan at the start of retirement, with blind adherence for the life, then a fixed stock/bond ratio performs better than an age-based ratio. In real life we don't blindly adhere to a plan, but reevaluate and adapt as we grow older.
-- We often should reduce equities as we age to lower retirement risk, but that decision should come from a broad-based reevaluation as circumstances change....and then only using guidelines with clear evidence to back them up. For example, there is evidence immediate annuities can reduce risk as we age. Another solid approach could be restarting/recalculating a withdrawal schedule at a later age using a shorter duration to derive a different stock/bond mix.
-- I doubt many here use an age-based formula, but not sure.
Thoughts....?