If you have a prudent financial plan, with no legacy desires, you very well might accidentally leave your heirs an estate large enough to have an estate-tax problem anyway! I’ll give the conclusion first (less technical) and then the detail of how it was derived (which is almost certainly more than you want to know).

If you use the 4% rule for spending, with a starting portfolio value of $1,000,000, in 30 years you have (in real, i.e. today’s, dollars):

- A 1/20 (5%) chance of running out of money
- A median (50th percentile) value of $2,000,000
- A 1/20 (5%) chance of $7,500,000 or more

Here’s how that was derived and some elaboration:

Using 1926-2020 data on stocks (CRSP 1-10), bonds (5YR TSY), and inflation (CPI) the real arithmetic mean was 8.8% for stocks with a standard deviation of 18.4% and 2.3% for bonds with a standard deviation of 4.7% (all annualized from monthly data by compounding the monthly return and by multiplying the volatility by the square root of 12). The correlation between stocks and bonds was 8.1%.

If we do projections using a Monte Carlo Simulation (MCS) with those figures using $1,000,000 as the starting value and $40,000 for an annual withdrawal (starting immediately) then this is exactly the 4% rule done with a MCS using historical returns to generate many more scenarios than we actually had historically. (Also, if you use rolling historical data, you over-sample the middle years and under-sample the beginning and ending years). I don’t have to inflation-adjust the withdrawals because I used real returns in the first place. Looking at the results at the 30-year horizon (again to match the Bengen research), the success rates range from 91% for all stock up to 97% for 40/60 (stock/bond) and then back down to 79% for all bond. I’ll focus on the typical 60/40 portfolio here. It has a success rate of 95%. I would say that validates the 4% rule pretty well – though expected real returns might be a little lower than historical realized real returns, this has no international diversification, no factor tilts, etc. but also no fees. Close enough probably to assume all of that roughly cancels out.

While this scenario runs out of money 5% of the time, in the median (middle) case there is just under $2,000,000 (and, to reiterate, all of this is stated in *real*, i.e. today’s, dollars). The 5% case (the 1-in-20 on good side) had an ending value of about $7,500,000, while the 95% case (the 1-in-20 on the other end) is zero.

Doubling everything so a hypothetical single client had a $2,000,000 portfolio to start with and withdrew $80,000/year (adjusted for inflation) to live on in retirement (and had no other assets), in 30 years there is an equal chance of zero and $15,000,000! The estate tax exemption is scheduled to revert to $6,000,000 (ish) and increase with inflation so we can compare our $15,000,000 to that $6,000,000 since both are real. $15,0000,000 minus $6,000,000 is an estate that would owe taxes on $9,000,000. At the current 40% rate, that is a $3,600,000 estate tax bill! To flog the deceased equine further, to have “only” a 5% chance of running out of money, there’s a 5% chance of pretty significant estate tax issues. (Of course in real life you see how it’s unfolding and adjust spending, gifting, etc. to ameliorate both extremes.)

That math is surprising to most people. We don’t generally realize how wide the range of financial outcomes in retirement can be.