Testing the 4% Rule Today
Recently, I’ve talked about the 4% rule and the sequence of returns risk that may put it at peril. Now it’s time to look at the 4% rule today. It’s certainly an interesting time to look at it given what’s going on in the world and how that impacts the markets. It’s always important to revisit certain items and see if they still hold up today.
After all, the original trinity study that produced the common 4% rule was done in 1998 and utilized data through 1995 adjusted for inflation. In addition to that, it only looked at a retirement period of 30 years.
While that’s likely enough if you’re retiring at 65, it may not be enough for those who want to retire early. On top of that, since 1995, we’ve added a few more years into the market and a longer data set is always bound to produce more accurate data as you’re covering more market conditions that may happen in the future. Plus, there’s more longer term periods to look at that may give insight into whether retiring with a 40 or even 50 year time frame is reasonable given certain withdrawal rates.
As such, I wanted to take a look at the 4% rule today with an analysis of data through the end of 2021. While the market turbulence since then has been worrisome, it hasn’t been anywhere close to enough to change the data much given that the S&P 500 is down maybe 12-13% since all time highs. If it continues to drop and we hit a recession and a prolonged market downturn then I might have to revisit this sooner than later but for now this data suffices. Let’s hope that doesn’t happen and the world can get back to being sane again soon.
Before we get into the data and this post will be very data heavy, please remember that the 4% rule isn’t going to be the answer for everyone. After all, it assumes static spending and withdrawals and that’s rarely the case for most people. There’s no flexibility in these calculations for a bond tent or a cash buffer to lower sequence of returns risk and no flexibility for lowered spending during a bear market or higher spending for emergencies. While testing and spreadsheets are useful for many things, they can never cover off on the many individual choices you’ll make or your own individual risk profile so keep that in mind.
What this is good for is a general sense of what may be possible and what makes sense.
While the below data may give you a starting point, it’s really up to each individual investor to do the math, account for their own personal strategies, spending patterns and risk profile. The data below reflects a basket of stocks and may not be applicable to someone with a more focused investing strategy. For example, while the recent market turbulence may not impact your retirement if you’re in an index fund, it certainly might if you’re 100% growth stocks.
With that in mind, the data below includes stock market returns from 1926-2021 and analyzes every starting point with the data set for the various payout periods shown. It captures all the best starting points and all the worst starting points to give a success rate for each payout period and withdrawal rate.
Anything at 100% means the withdrawal rate has never failed for the respective payout period and anything below 100% means there’s been at least some failures within that time frame.
It is inflation adjusted using historical CPI, assumes a total market stock portfolio with a .08% expense ratio and ten year bonds for the bond allocation.
With that in mind, let’s take a look at the data.
The 4% Rule Today and Across Longer Withdrawal Periods
I’ll let you stare at the wall of grey and analyze the data as much as you want but I’ll cover off on the important highlights that stick out to me. There may be other things to discuss but it’s important to take some time to digest the data as well before we do that. Leave comments below if anything in particular sticks out to you that I may not have covered.
3% as a withdrawal rate seems to be a lock based on historical data as long as you have at least half your money in a total market stock fund. It’s the maximum you can withdraw in this test that has a full guarantee across 50 years.
Naturally, the success rate starts to drop off in longer periods as the withdrawal rate goes up. Given that, does the 4% rule still hold up?
Well, the success rate with a 30 year withdrawal period and at least a 50/50 split does drop from 95-98% depending on allocation as of the original test to 89-92% as of this test. However, that’s still pretty favorable.
It basically means that if you are a robot who just follows this rule blindly, you have a pretty good chance of success and minor optimization can shoot that up higher. The 4% rule never said that it was a lock solid guaranteed success(as it wasn’t 100% on all fronts) but that it was almost a certainty given historical data.
This test puts a tiny wrinkle in the original results by lowering the chance of success a bit but it still remains pretty good. The 3.5% numbers these days do match up a bit better with that historical back test and might be a better comparison right now. Maybe it’s the 3.5% rule now!
HOWEVER, one thing to remember, is that the original rule was always measured by a 30 year withdrawal period and success does drop off quite a bit beyond that especially as your stock allocation drops. The 4% rule with a 50/50 allocation only has a 50% success rate when tested against a 50 year withdrawal period. That’s a pretty big difference and makes the 4% rule as a rule a bit more risky given a much longer withdrawal period. A 75/25 allocation does have a much higher success chance though but still only sits at 74%.
On the value of stock allocation, another item that stands out is that while bonds do help with success rate in shorter time periods, their usefulness is less impactful the longer you depend on the money. For, example at a 4% withdrawal rate, the 30 year success rate of a 50/50 portfolio beats the one of 100% stocks by 3%. However, once you get to that 50 year mark, the 50/50 split trails by a whopping 24%.
The reason for that is that stock volatility gets smoothed out across a much longer time period and the low returns of bonds add less value when that happens. Add inflation, volatility and all that stuff and longer time periods benefit a lot more from high stock returns as long as you can stomach the few downturns that are bound to happen within that time period.
While I did do testing beyond the 4% rule, it’s clear that once you get far beyond that, the success rates start to get a bit worrisome especially if you want to depend on that money for an extended period of time.
A lot of that is due to simple math. If long term stock returns are 7% after inflation then you probably can’t withdraw 9% adjusted for inflation and expect to stay above $0 outside of really favorable investing periods. While success rates at those high levels do happen, they are more rare and difficult to depend on.
Another key part here is that while a 100% success rate might seem like a must, it actually isn’t. If you look for a 100% success rate in anything, that is a very risk averse way of approaching a withdrawal period. It might be good for some people who are that risk averse and want to avoid ANY stress when it comes to investing but it’s often not reality.
At the end of the day, you want to get the best combination of safety while also maintaining a reasonable withdrawal rate and spending amount to support your lifestyle. Even the 4% rule in the original study didn’t have a 100% success rate but was close enough so that it worked. If you look at 30 year periods here, then the 4% rule still applies in that sense with a tiny bit more risk.
After all, a 90% success rate is often probably good enough considering you can cut your spending in years where the market is down and may not actually need the money for the full 50 years(social security might come into play, you may have another source of income or inheritance, you may pass on earlier than that, etc.). Like I mentioned before, the 4% rule is a very static measurement and doesn’t take into account all those other factors, especially variability in spending, other income sources or even strategies that may mitigate sequence of returns risk.
If you’re too conservative, you’ll end up maintaining a lower than necessary quality of life and be left with much more than you want to be left with when you pass on. Unless you’re planning to leave a massive estate to your children or to charity, it’s important to have a good balance between what you choose to withdraw and the risk you’re willing to take on.
In most of these high success rate scenarios, the investor is left over with much more than the starting amount. For example, the average end point of someone starting with $1M with a 3.5% withdrawal rate and a 50 year retirement period is $8.7M. That’s despite an 89% success rate.
The important thing to remember is that while 11% of the simulations fail, they fail in the worst possible scenarios only when there’s a massive downturn in year 1 or 2 or a prolonged down market.
In most scenarios, that doesn’t happen and a 3.5% withdrawal rate ends up growing your portfolio a lot which leads to that high average total.
That may leave a decent margin of safety outside of the really terrible periods where you have very bad luck with sequence of returns risk.
Even in those scenarios, setting up a bond tent you can draw on if the first few years are rough or starting with a years worth of cash may shoot those 80-90% success rates right into the 100% range pretty quickly but that’s something to test for another day.
For now, and based on historical data, it’s pretty clear that at least a 50% allocation to stocks is the way to go for shorter time periods and at least 75% stock is the way to go for longer time periods. The 3% withdrawal rate is super safe with those parameters and 3.5-4% is pretty safe for 30 year periods which is in line with the original test.
However, 4% does get a bit more risky as you move up and into the 50 year withdrawal timeline.
For me, I think somewhere in that 3.5-4% range will be just about the right number. I know I can employ some strategies that will make periods where sequence risk is a concern a bit less problematic and always want to aim for longer periods than shorter since you never want to run out of money well into your retirement journey.
On top of that, once you get past those first few years where sequence risk is much more pronounced, that 3.5-4% on a growing portfolio may mean your income is higher than your spending and you can cut back. After all, if you start with $1M and get lucky with returns and are sitting on $2M in a few years, you can probably still depend on similar income and can reduce that withdrawal rate to add more margin of safety. Or you can go hog wild and spend double, it’s up to you!
It’s the opposite end where it starts to hurt.
Years like this one will always happen and while the stock market drop so far isn’t big enough or long enough to impact any of these numbers(unless you were all in on riskier growth stocks), it certainly puts into mind the type of markets that drive failures in these types of back tests. If you start with $1M and experience some bad years and are left with $500k after a few years, then you might be in trouble.
It’s important to think of both sides of the coin and that’s what this type of back test allows you to do. Yes 4% is generally pretty safe but maybe in the worst scenarios it might not be so perhaps you need to be prepared to do certain things when that happens.
There will be tragedies like the one we see this year and have seen in the past that push market returns down and may keep them down for an extended period of time and it’s important to be ready for those times whether they happen today or in future years.
Backtesting isn’t a perfect science nor does it give you the 100% correct answer but it does give you a starting point from which you can proceed and hopefully this post did that for you.
It’s probably something I’ll run every few years to get an idea of where historical market returns are pointing because you never know what kind of market is ahead and it’s important to look at history as a guide of what could but doesn’t necessarily have to happen.