Halls of Shrewd'm / US Policy❤
No. of Recommendations: 2
In a few years time, probably 2028, I will begin taking withdrawals from a retirement account.
The account is currently 100% Berkshire.
The withdrawals will be fixed (SEPP).
And because I can't control the price of Berkshire, I wanted to "work forward" and build a cash cushion.
For example, say the portfolio is $1m and the withdrawal $60k/yr.
The plan was to sell 5% of shares per year, spread across 4 quarters, starting now'ish in preparation for that initial $60k withdrawal in 2028.
But now that I'm typing this, maybe I should just sell $60k of Berkshire once per year starting in 2028 and be done with it?
That sounds the simplest.
P.S. I've studied Jim's annuitization model, and was originally planning to use that, but changed my mind to sell 5% of shares per year because it's a tad simpler and what Buffett recommends. I don't know.
No. of Recommendations: 4
In a few years time, probably 2028, I will begin taking withdrawals from a retirement account.
The account is currently 100% Berkshire.
The withdrawals will be fixed (SEPP).
And because I can't control the price of Berkshire, I wanted to "work forward" and build a cash cushion.
For example, say the portfolio is $1m and the withdrawal $60k/yr.
There is nothing special about having only BRK in a retirement account, other than it is very non-diversified and thus has more risk.
Taking 6% withdrawal is well above the "save" SWR of 4%. You may want to take this discussion over to the Retirement Investing board.
Two other comments:
1) You can't control the price of any investment, not Berkshire and not the S&P 500 whole market. BRK isn't anything different here.
2) Statistically a cash cushion, also called a cash bucket allocation, is worse than just keeping all the money invested.
No. of Recommendations: 3
So I used the tool Adrian mentioned at
https://testfol.io to backtest a $60,000 fixed annual withdrawal (not inflation adjusted) from a 100% Berkshire portfolio.
After 10 years (7/1/2015 - 7/1/2025) withdrew $600,000 end with $2,268,557.67.
After 20 years (7/1/2005 - 7/1/2025) withdrew $1,200,000 end with $3,863,737.32.
After 30 years (7/1/1995 - 7/1/2025) withdrew $1,800,000 end with $9,637,405.03.
I guess that makes sense. You're more likely to end up with way more money in the end.
No. of Recommendations: 2
I guess that makes sense. You're more likely to end up with way more money in the end.
That is true with the 4% safe withdrawal strategy as well. Of course that is true in the back tests and you unfortunately never know what will happen in the future.
I have a conservative portfolio but am happy to see after a couple of years of retirement to see my portfolio is larger (not by a lot) than when I retired despite relying on it for about 70+% of my expenses.
Rich
No. of Recommendations: 4
So I used the tool Adrian mentioned at https://testfol.io to backtest a $60,000 fixed annual withdrawal (not inflation adjusted) from a 100% Berkshire portfolio. This is interesting but not very useful. The purchasing power of a fixed dollar amount over even 10 years is much lower. Over 30 years, it's horrible. Assuming you need the funds to pay for living expenses, you have to make allowance for the effect of inflation.
No. of Recommendations: 10
I think the best withdrawal strategy is the RMD as required from retirement accounts by the IRS. Each year you withdraw a percentage of your account's latest balance, calculated based on your life expectancy. So you'd withdraw about 3.6% at age 70, gradually rising to 10% at age 92, and 20% at 104 if you should live so long.
Note that the IRS tables are more conservative than people's true life expectancy. Actuarial tables show a life expectancy of 17 years at age 65 for men and 20 years for women. So you'd be starting with 5%+ withdrawals at age 65. The IRS mandated RMD is only about half of the true life expectancy.
Elan
No. of Recommendations: 9
that is true in the back tests and you unfortunately never know what will happen in the future
A much better strategy for people selling a portion of BRK for living expenses is to sell slightly more than needed and build a cash buffer when the stock is selling at the higher end (or if you are lucky, more than) of intrinsic value as it had done in the first 4 months of 2025. And sell less or ideally none at all when the stock is very cheap. When the stock trades around 1.5 time books (or sort of a mid point of IV range), one can sell a "normal" amount.
No. of Recommendations: 2
Yes, I’d imagine $60K of current value goods & services in 20 years would cost one >$96K, likely More given threat of inflationary risks long term.
Warren’s net worth ($142B) appears up 210% vs. his net worth in 2006 ($46B when he announced the huge Gates Foundation gifting), despite his 4-5% annual giving.
Berkshire share prices have held up pretty well in their growth and purchasing power, however I doubt we will do as well over the next 20 years given our $>1T size. Inflation + 6-8% would be nice.
No. of Recommendations: 16
Note that the IRS tables are more conservative than people's true life expectancy. Actuarial tables show a life expectancy of 17 years at age 65 for men and 20 years for women. So you'd be starting with 5%+ withdrawals at age 65. The IRS mandated RMD is only about half of the true life expectancy.
Elan
It's even worse than that. Life tables look at the life expectancy of ALL people at a given age (say, all males 65 years old) and tell you how long they will live, on average. But this includes everyone, including people literally on their deathbeds, and should not be used to calculate the life expectancy of a 65 year-old who has no known serious risk factor or life-threatening illness.
For instance, in a group of 100,000 65 year old men living in the USA, life table calculations may tell us that they will live for an average of 17 more years. For the sake of illustration, let's say these 100,000 people include 1,000 people with a serious, advanced disease, like metastatic cancer, let's say with an average life expectancy of 2 years and 39,000 people without such a diagnosis but with a serious risk factor for death (diabetes, smoking, aortic aneurysm, emphysema, whatever), with an average life expectancy of 8 years. Then it is easy to calculate that in the 60,000 people without those serious risk factors, average life expectancy will be (100,000*17-1000*2-39,000*8)/60,000 = 23 years of remaining life.
Of course this is an oversimplification - there all all sorts of intermediate risk groups, not just these 3 broad groups. But the point is that if you are a relatively healthy 65 year old, your life expectancy will be much longer than the average for ALL 65 year-olds. When you are planning how much you can safely remove from your savings each year, this can make an important difference.
Regards, DTB
No. of Recommendations: 7
It's even worse than that.
65 year old men living in the USA, life table calculations may tell us that they will live for an average of 17 more years.
Then it is easy to calculate that in the 60,000 people without those serious risk factors, average life expectancy will be (100,000*17-1000*2-39,000*8)/60,000 = 23 years of remaining life.
Well, it's actually not worse than that. The IRS table starts at a 27 year life expectancy at age 72, while the statistical life expectancy at that age is 14 years for men and 16 for women.
Elan
No. of Recommendations: 14
I'm not familiar with the intricacies of the SEPP rules, but I gather the key question is: given that you need to have sold a specific number of dollars of stock by a specific future date, when should you do that?
My answer would be "at any time that the most likely return from the stock until that future date is negative"
Make a few assumptions. They don't have to be perfect.
Let's say you use these:
* Assume inflation of 2.75%, a pretty typical MCT inflation figure for the last 2 years.
* Berkshire's value per share can be approximated as a multiple of book per share.
* Both will grow at around inflation + 7%/year in the next few years on trend. It has been inflation + 8% for ages, but one can assume a bit of moderation.
* Future price will average 1.43 times known book per share. That's just a hair higher than the 20 year average of 1.4.
* June 2025 book per share will come in around $465000, equating to $310 per B.
* Interest on cash will be just enough to cancel inflation.
That set of assumptions gives you a table something like this (real prices meaning in today's dollars, nominal meaning the apparent price at the future date shown):
real book real p nominal
2025 $ 2025 $ price
Mid 2025 310.00 443.30 443.30
Mid 2026 331.70 474.33 487.38
Mid 2027 354.92 507.53 535.83
Mid 2028 379.76 543.06 589.11
Mid 2029 406.35 581.08 647.68
So, round numbers, buy and hold might get you around $648 per B share in 2029. So, if you get an opportunity *any time before then* to sell for that much or more, feel free to sell then sit on cash till then, especially if it's materially more. You can revise the table as new book-per-share figures come in, but it probably won't make a huge difference.
This won't give an optimal solution, but it will give an eminently sensible solution. If the stock price is above what you might reasonably expect at your target future date, it's not crazy to sell then wait.
For bonus points: using a SEPP, I think you have this problem every year 2029 and later?
So you could have a separate table for each future year. Sell each year's "chunk" as soon as the reasonably expected return until that future time is negative. So, during a real price pop like we have just seen, before or after 2029, you might end up selling a few years' worth of withdrawals.
Jim
No. of Recommendations: 5
Thank you Jim.
I'm not familiar with the intricacies of the SEPP rules, but I gather the key question is: given that you need to have sold a specific number of dollars of stock by a specific future date, when should you do that?
For bonus points: using a SEPP, I think you have this problem every year 2029 and later? So you could have a separate table for each future year.
Yes, you nailed it! A SEPP (IRS rule 72t) allows me to withdraw a fixed amount from a tax-deferred retirement account (IRA) before age 59.5 *without penalty*. In my case, I will need to withdraw the fixed amount for 10 consecutive years.
Cool, I will build those tables.
This won't give an optimal solution, but it will give an eminently sensible solution.
Yeah, I don't know what the optimal solution is. I guess it depends on what you're optimizing for. There are just tradeoffs.
I considered:
1) selling the fixed amount once per year; easiest, but likely leaving money on the table, because not a valuation-based decision
2) starting a few years early and selling quarterly to build cash; easy, and would get average prices
2) selling the 10 years worth of withdrawals at once; likely leaving the most money on the table, but sleeping well at night
3) the annuitization method, essentially the same as the tables using a P/B multiplier
Thanks again!
No. of Recommendations: 7
If you do use a "2029 price" target, bear in mind that you might not want to check the situation too frequently.
Why so?
If you check daily, you will sell at the very first moment that the estimated forward return goes negative, and never get any extremely high prices for what you're selling.
If you check only every (say) month or two, there's a decent chance that some of your sales will be at levels above your target because prices kept rising until you look at them. Most stretches of high valuation tend to last for at least a couple/few months, so you're unlikely to completely miss an opportunity. And of course it's much less work! For the same reason, I wouldn't just put in a good-till-cancelled sell order for the target price.
Another thought:
On the theory of "minimize maximum regret", you could also use this strategy, but sell only half of your future need the first time an opportunity arises, then wait for another opportunity a minimum of X months later before checking prices again and considering selling the second half. One of the two is certain to get you a better price than the other. If it's the first one, you'll be glad you sold the first half when you did. If it's the second one, you'll be glad you waited to sell the second half. Either way you feel like a genius for not leaving money on the table : ) The disadvantage is that there may be only a single stretch of rich valuations in the next few years--who knows what Mr Market will feel like? But it's not exactly fatal...you'd just sell the last chunk at the last minute, which is what you were probably figuring on doing anyway.
The overall strategy isn't specific to SEPP rules - having one target price for each future period could make sense for *anyone* living from a portfolio of periodically liquidated Berkshire stock. If you need $X000 to live in 2031, sell $X000 worth of stock as soon as the market price is above what makes sense as the best guess for 2031. Sit on suitable-term TIPS till then. If most of your sales are done at prices higher than the "likely" price at the date of future expenditure, it seems reasonable to expect that you're increasing your average realized price quite a bit. You should have a lot fewer instances of having to sell a a chunk at a temporarily low price because that's all Mr Market happens to be offering on January 1 that year.
Jim
No. of Recommendations: 12
Of course this is an oversimplification - there all all sorts of intermediate risk groups, not just these 3 broad groups. But the point is that if you are a relatively healthy 65 year old, your life expectancy will be much longer than the average for ALL 65 year-olds. When you are planning how much you can safely remove from your savings each year, this can make an important difference.
Oh there’s an easy one that predicts life expectancy differences of 20% or more of adult outcomes: whether you are a knowledge worker or a blue collar worker:
“Studies indicate a disparity in life expectancy between knowledge workers and blue-collar workers, with knowledge workers generally having a longer lifespan.
Here's a breakdown of the findings:
Overall Life Expectancy: Mean total life expectancy is reported to be highest among executives and managers at 73.2 years, followed by clerical workers (also knowledge workers) at 72.0 years. Unskilled blue-collar workers are found to have the lowest mean total life expectancy at 63.65 years.”
I’m guessing that the calculations are close or similar until the person reaches adulthood (probably not, actually, given poverty, localized crime rates, etc.) but once age 20 is reached, the outcome changes pretty dramatically: from 43 years for grunt work to 53 years for think work.
That is quite a statistically different outcome.
No. of Recommendations: 0
Oh there’s an easy one that predicts life expectancy differences of 20% or more of adult outcomes: whether you are a knowledge worker or a blue collar worker:
I suspect the fraction of blue collar workers with significant retirement accounts to draw down is quite small to begin with. Would be interesting to see how much of life expectancy reduction is due to dangerous jobs vs. lifestyle (less access to good food, good medical care, etc. for blue collar workers).
