No. of Recommendations: 14
I too was struck by those posts...so I saved them. Here they are!
From time to time I ponder the interesting statistico-arithmetic problem of funding a retirement of unknown length with a stock portfolio.
A while back, about 5.5 years ago, I mooted a method for someone who found themselves holding nothing but Berkshire stock.
https://boards.fool.com/sorta-ot-swr-on-brk-319408...The question it addressed:
If one is willing to put up with some variability of income, what's the most you could cash in per year or per quarter, with zero risk of ever running down your portfolio?
It starts with the observation that ultimately it's the increase in value of your portfolio over time that determines how much you can spend in retirement.
So, just estimate the value of your portfolio periodically, and liquidate only the amount that leaves its real inflation-adjusted value constant.
If you get that right, then by definition your stream of liquidations will last forever, or at least as long as the value of the shares is rising.
The specific proposal was to value Berkshire shares regularly, smooth that with a four year average,
and sell only the amount that leaves the value of your portfolio (using that smoothed valuation number) unchanged.
I like to use a weighted moving average for such calculations (WMA) rather than a simple average, as it considers more recent information to be more important.
A weight of 4 on the most recent year, 3 for a year ago, 2 for two years ago, 1 for three years ago, and divide the sum by 10.
You can do the same ramping thing with quarterly figures by using weights 1-16 and dividing the sum by 136, which is what I did here.
For the valuation metric I used the average of 1.5 times book per share and my "two and a half column" metric, both done quarterly.
[I did add one quirk: I assume that any drop in a valuation metric larger than 2% will be transient and can be ignored.
So each quarter, I use the higher of "current value measure" and "98% of peak to date".]
All figures adjusted for inflation. So, all of the following figures are inflation adjusted into today's dollars.
So, imagine one had taken this suggestion to heart when the post was made, and you owned a million bucks worth of Berkshire stock (in today's dollars), and retired that day.
Not a million at market price, but a million in intrinsic value based on the smoothed valuation metric.
Today, by construction, your shares are still worth the same million bucks. The market price is not quite that figure, but that's what they're still worth based on this quarter's estimate.
Of course, each liquidation leaves the value unchanged, but is of varying size for multiple reasons.
Partly because the growth in value per share is irregular, and partly because you have to sell at market prices which are themselves irregular.
Anyway, here are the figures since the post, in today's money:
Price Value
2015 liquidations: 98071 11.0% 9.8%
2016 liquidations: 76205 8.6% 7.6%
2017 liquidations: 79813 8.0% 8.0%
2018 liquidations: 91997 8.9% 9.2%
2019 liquidations: 94093 9.5% 9.4%
2020 liquidations: 78705 9.2% 7.9%
i.e., you could have sold stock quarterly adding up to that much (in today's dollars--inflation protected) and still have a portfolio worth as much as it was in 2015.
The first percentage is the real value of the liquidated shares as a percentage of the average portfolio market value in the prior year.
The second percentage is the real value of the liquidated shares as a percentage of the constant $1m intrinsic value of the portfolio.
Of course, we started out with the requirement that you live with variability of the amount of cash raised.
Some years are not so kind, notably around a big recession when not much value progress is visible.
Note the dip 2009-2010:
2004 liquidations: 90410 10.4% 9.0%
2005 liquidations: 81310 10.6% 8.1%
2006 liquidations: 79693 9.9% 8.0%
2007 liquidations: 97157 10.6% 9.7%
2008 liquidations: 74746 8.7% 7.5%
2009 liquidations: 32368 5.2% 3.2%
2010 liquidations: 35341 4.5% 3.5%
2011 liquidations: 41021 5.8% 4.1%
2012 liquidations: 57879 7.8% 5.8%
2013 liquidations: 91798 10.2% 9.2%
2014 liquidations: 113278 11.5% 11.3%
Without a doubt, that's a big cut in your income around the credit crunch.
But, to keep it in perspective, it only dipped down to the sort of SWR that would be considered aggressive using SPY these days.
Most of the time the cash generated has been very impressive, to say the least.
In the last 58 quarters it has paid out a million bucks (in real terms) starting with a million bucks (in real terms) and no compounding was allowed.
That's almost twice the total real income you'd have received from one "standard" SWR strategy--4% of initial portfolio value per year, adjusted for inflation.
And with vastly lower risk of outliving the nest egg.
As always, the sustainable rate comes down to the rate of rise in observable value of the asset you've chosen, plus or minus a one-time adjustment for its valuation level when you start.
So long as a share of Berkshire keeps rising in value at over (say) inflation+6%/year, then one can cash cash out that much on average without ever reducing the value of the portfolio.
The method proposed automatically cuts the payouts if/when business results falter, though with a
slight lag, so any slowdown in the business becomes a slowdown in withdrawals, not a risk of eating into the capital.
Jim
For anybody interested in playing with it, these are the quarterly valuation metric numbers I used, from end 1989 to end 2020.
All in today's dollars with CPI at 260.23
Early years are a little more rough and ready, take them with a grain of salt.
For one thing they use only book value, not the average of book and my two-and-a-half-column metric.
13393 (end 1989)
13125
14145
13863
13863
15412
15788
17030
18362
17994
18975
20536
21451
22207
22588
23567
23880
23402
24279
26792
26490
29079
31618
34143
35918
38551
41517
43764
47153
48384
55854
55517
62081
68083
71868
70430
90670
91101
90389
89279
89279
88581
87494
88632
91377
89550
89550
89550
87759
87759
88356
87463
88241
92715
99914
101204
114447
114284
113897
113940
117626
117574
118167
118842
120406
124363
124867
129868
138954
139478
142811
148530
147950
145717
145559
144991
142803
142648
142091
145121
147518
154027
150947
155376
165652
169289
168560
165903
172935
184249
183918
190855
196100
205443
209172
214026
228295
230255
235023
242052
245433
250783
253003
253663
259508
258542
259019
261674
273284
283963
289706
294525
315042
314335
323680
341445
336573
344452
354710
369120
386097
378375
378375
382341
409719 (end 2020)
If a person retiring today had nothing but a big block of Berkshire shares,
what's a safe withdrawal rate for income to last through retirement?
I don't particularly recommend starting from this scenario, but it's an interesting problem.
Here's yet one more possible solution.
Both higher income and more guaranteed sustainability, at the expense of some variability:
Each year, estimate the value of a Berkshire share.
Maybe classic two column, or something like my 2.5 column model, or just 1.5 times book/share.
Each year, sell enough shares so that your remaining shareholding is
worth (after inflation) the same amount as its peak to date.
Not market price "worth", but intrinsic value "worth".
(peak will be almost, but not always, equal to the same value every year: there will be the occasional down year in valuation).
In effect you will liquidate the amount of all your after-inflation gains, but only that amount.
By construction, the real value of your shareholding will never change over time, and you can't ever run out of money.
Your income from year to year will vary with the firm's fortunes, but not really all *that* much.
Keep a cash cushion, as there will be the occasional down year in
apparent value so this approach will recommend no sales some years.
Now, holding a year or two of cash is the easiest fix to deal with variability, but a fancier
fix for that is to replace each year's share valuation with the four year average.
Sell whatever amount leaves the value of your shareholding (based on four year
average of valuations) at the historical peak of that four year average in the past.
For a sense of the smoothing effect, the four-year-average inflation-adjusted 2.5 column
method valuation has not risen less than +2.3% in any of the last 20 years,
even though without averaging it dropped by -10.7% in the credit crunch.
Four-year-average inflation adjusted valuation has risen 10.5% to 12.3% in the last three years.
Sure, it'll vary, but something like "2% to 12%" is an attractive SWR given that you're guaranteed not to go broke.
Other than the disadvantage of variability--
- Any retirement approach designed to last forever is definitely going
to last longer than you do; you never get to spend the capital.
- By extension, it requires that you have a large portfolio.
It has to be big enough that you can live on just the income.
Obviously this approach works for any type of portfolio which lends
itself to a simple numerical valuation method that's "good enough".
It emphasizes the point that it's the increase in value of your
portfolio over time that determines how much you can spend in retirement.
Jim
