FWIW, my valuation comes up with these figures.
I never attempt to get the right multiples, as I’m interested only in the growth rate of the
metric and what the usual ratio of market price to metric is.
So, think of these figures as [some unknown constant] times a reasonable estimate of intrinsic value.
These figures are all inflation adjusted.
2014Q4 **$288063**
2015Q1 292721
2015Q2 298128
2015Q3 300786
2015Q4 **304761**
2016Q1 287421
2016Q2 288676
2016Q3 299216
2016Q4 **312624**
2017Q1 324456
2017Q2 329789
2017Q3 333254
2017Q4 **343624**
2018Q1 347075
2018Q2 361551
2018Q3 382583
2018Q4 **377936**
2019Q1 393950
2019Q2 402792
2019Q3 418911
2019Q4 **435580**
2020Q1 403706
2020Q2 419327
2020Q3 428648
2020Q4 **455364**
2021Q1 475611
2021Q2 507054
2021Q3 520747
2021Q4 **530566**
2 year rate of change of observable value: inflation + 10.4%/year
3 year rate of change of observable value: inflation + 12.0%/year
4 year rate of change of observable value: inflation + 11.5%/year
5 year rate of change of observable value: inflation + 11.2%/year
6 year rate of change of observable value: inflation + 9.7%/year
7 year rate of change of observable value: inflation + 9.1%/year
Oddly, this series is consistent with value rising more quickly, not more slowly, over time.
Note, I apply a substantial haircut to the value of the equity portfolio.
For each period, I have capped the valuation of big positions to a finite multiple of earnings. (chosen to be optimistic but not exuberant)
In effect, I value big positions based on the trajectory of their earnings rather than the trajectory of market prices.
For year end 2021, my haircut was $55.1 bn, or $37300 per share, biggest to date.
78% of the adjustment was for Apple. 13% of it was for Coke and Moody’s.
There is one purely arbitrary patch to the data: the one horrible quarter of earnings 2020-Q2 was “patched” by eyeball to be on trend.
It was clear that the bad earnings wouldn’t last, so I didn’t consider it a good indication of true value.
In the table above, the 2020-Q2 operating earnings were replaced with the average of the 2019-Q2 and 2021-Q2 figures.
Based on the historical ratio of this metric to market price, one might expect one year returns in the range of inflation + 5.1%.
Depending on which model I use, it gives figures ranging from inflation + 2.5% to inflation + 8.1%.
So, if inflation comes in at a mere 3%, that would give a target price end Feb next year end around $519100 (a bit over $346 per B).
The lower forecasts generally look only at market valuation multiples 2008 and later.
The higher figures tend to come from models that also consider earlier data, generally 2003 and later.
Some models use this valuation metric (generally the lower forecasts), and some use book value.
The book value models are more optimistic lately because there is no valuation haircut on big expensive stock positions.
To recap how my value metric is calculated:
- A multiple of 15 on “steady things”: after tax earnings from rails, utilities, manufacturing/service/retail, and a cyclically adjusted estimate of underwriting profit.
- Investments per share at market value, less a haircut for overvalued positions.
- Minus a figure, proportional to the size of the insurance operation, to count the fact that some
portion of the portfolio will always be earning nothing and is therefore worth nothing to us.
That’s estimated as 30% of float. (not intended as an estimate of the size of the cash pile–that’s something subtly different)
If you crunch these numbers, you’ll see there is no trend in the “fair” P/B ratio rising or falling.
In other words, the valuation level “P/B 1.5” means about the same thing it did quite a few years ago.
What you might call a reasonably rare but reasonably fair multiple.
Presumably there will be some divergence some day, but not yet.
Jim