The failure of the 4 % withdrawal rule
A general investment rule says that if you invest your savings in the stock market, you can safely withdraw 4% of the savings each year, with very little risk that you savings go to zero.
Otherwise stated, if you save 25 times what you need each year, you have enough savings for the rest of your life. The returns of the stock market will more than compensate your withdrawals.
The theory is based on “The Trinity study”, which looked at the past (1926-1995) performance of the US stock market and came to this conclusion:
Using a portfolio of 100% stocks, in 94 % of the simulated cases, you survive at least 30 years without your savings becoming depleted. For the nerds, this was the Trinity Study table with probabilities using 100 % stocks:
There are also tables for a portfolio mix stocks / bonds, but given the extremely low returns for bonds today, it would be totally nuts to rely on these numbers to predict future returns. So let’s further focus on stocks only.
The Past = The Future?
The Trinity study was updated in 2018, using data going to 2017. See here.
Survival rates given here: a 100 % stocks portfolio now still survives 30 years for 94 %; but the survival rate for 40 years is 89%.
More importantly, the author rightfully says: ” Again, it is important to be clear that these success rates are based on US history. It is faulty logic to think that these are the success rates applying to new retirees today. The particular situation today is that interest rates are so low relative to history, and this is a very important matter when assessing the viability of different withdrawal strategies. “
The stock market return can in a simplified model be reduced to growth of the economy + inflation + dividend yield. These were the factors in the past (1950 – 2019):
To obtain the stock market return from 1950 to 2019, you have to include one more factor: the stock market was extremely cheap in 1950: a Shiller CAPE of around 11. And it is very expensive in 2019: a CAPE of 25 or more. To account for that 140 % increase over 70 years, I add 0.5 % of annual returns.
Conclusion of 1950 – 2019 US S&P500 stock market index
Even without statistics or simulations you can see that there is a large probability that with a stock market return of 7.2 % ABOVE INFLATION, and if you take out 4% each year, you will end up more years with profits than years with losses. After all, you are making an almost double amount of profits (7.2 %) compared to the amount you take away (4 %). Obviously, there are also bad stock market years, but most of the time you will have increasing savings.
The Past = The Future? (2)
Now my question is: given a different reality in growth, inflation and dividend yields, and given that I am living in the reality of taxes and an index fund expense of 0.25 %, is this 4 % rule still valid?
Many FIRE enthusiasts reply with the simplistic answer: “It has worked in the past for a very long time, so it will work in the future”. Others simulate the trinity study by re-using the same numbers from the past, with an update of the last decennium . They then conclude that the stats still work. However, they use the same 70 year- period for their backtesting, which occurred under different circumstances that the current situation.
Growth, inflation and dividend yield are very different today. You can hardly argue that the average S&P 500 dividend yield today is 7.44 % as it was in 1950, can you (see here for historical S&P 500 dividend yield data)? (Hint: in 2019 is was 1.78 %)
Or that the annual inflation today is 7.88 % as it was in 1951, can you (see here for historical inflation data for the US)? (Hint: in 2018 is was 2.44 %)
You find hundreds of websites telling you the magnificent story of the 4% rule, spreadsheets, calculators (nice example here, but it states ” using stock market history … to give you the information to judge your savings…” ), … however they all assume the past as a perfect predictor for the future?!
But if the economy today is vastly different, and the economic basic numbers are very different, why should you then use the stock market returns of these years as a basis for your simulations for the future? Because you like the outcome? That is distortion of the facts to confirm your prejudice!
There are many more reasons to show that the past is not the future. You can find a number of them here. The most shocking statement is this: ” Using 109 years of data for each of 17 different developed countries, Pfau (author of the updated Trinity study) determined that a 4% withdrawal rate with a fixed 50/50 asset allocation would have failed in all 17 countries. Yes, a 100% failure rate “. Now please explain to me why you would choose the stock market returns of the uniquely exceptionally best stock market (USA) during its most exceptional period in history as your basis for an extrapolation of future returns, and why you would ignore 17 other civilized, modern, western economies?
Same thinking here: ” Since this performance is unlikely to be repeated, it casts doubt on the 4% number “
Another shortcoming in the Trinity study is that there is an assumption of zero taxes and zero expenses. Expenses on today’s index funds may not be high but they are not zero. Also, purchasing involves expenses and selling involves expenses (and you do want to withdraw your 4 % each year, don’t you?), and taxes are far from zero. Somehow, all websites citing the magnificent 4% rule ignore all of thoses expenses also, probably for simplicity reasons.
In Belgium we have at the moment of writing (december 2019) a high tax on dividends (30 %). And since we cannot buy the cheapest ETF’s of the world (Vanguard), at a TER (Total Expense Ratio) of less than 0.1 %, we also have to adapt to reality. Most Euopean index trackers are around 0.25 % TER or even higher. On top of that, there is a purchase fee (at least in Belgium), and in some cases a sales tax (1.32 %). I ignore these latter fees because they are a one-time event and 1.32 % on the 4% which I withdraw, so technically speaking I should simulate a withdrawal of 4.05 % (4 % + 1.32 % on 4 %).
Future expectations – USA
So to simulate the future, we should come with realistic expectations. I do not have a crystal ball and neither have you, but let’s assume that the current situation is closer to reality that the data of 70 years ago. So I take as assumptions:
- Economic growth: 2.1 % (See here and here)
- Inflation: 2%
- Dividend yield: 2% (see here for latest years)
On top of that, we are currently looking at an overvalued, CAPE = 25 stock market level. On the short term, it could go anywhere, but on the long term, certainly if interest rates or inflation would normalize, the PE ratio would rather go down than up, so I include a -0.5 % annual correction for the future.
This leads to this table:
4 % SAFE withdrawal rule??
The Trinity study simulated a 4 % withdrawal in a 7.2 % stock market (excluding expenses).
Today we are facing a 2.75 % stock market return (including expenses).
I do not need simulations to show me that a 4% withdrawal has little chance to survive a 2.75% stock market. Also a 3.5 % withdrawal is risky, and even a 3 % withdrawal may end up frequently in an extinguished savings account. You probably have to go to a 2.5 % withdrawal rate to have some chance of surviving in your old age.
After I have run the statistics I will update with more ‘accurate’ numbers, but every trial to run stats is always dependent on the input you provide for expected stock market returns and inflation environment. And all statistics and simulations I have seen so far use the past returns as unrealistic input for the future.
For completeness, I also give the prediction for the European stock market. Growth is expected even lower in the ‘old continent’, but dividend yields are currently much higher, and the market is not overvalued on average.