Avoid making negative or unhelpful posts, and instead focus on providing constructive feedback and ideas that can help to move the discussion forward.
- Manlobbi
Stocks A to Z / Stocks B / Berkshire Hathaway (BRK.A)
No. of Recommendations: 24
Someone asked for opinions about when to buy [back] in to Berkshire.
I thought this image might be helpful
http://stonewellfunds.com/SmoothedRealValuePerShar...A long time ago I proposed that the ultimate safe withdrawal rate is to liquidate no more than the amount the value of the shares has increased. Prices might go up and down, but the real value of your holding would never change. You could withdraw money forever, or at least as long as the real value kept rising. Since valuation metrics are themselves pretty volatile, I proposed using a smoothing method. At worst it would be an over- or under-estimate for 2-3 years.
The smoothing proposed was a 16-quarter weighted moving average of real book per share. I also use a 16-quarter weighted moving average of my two-and-a-half column valuation figure. Once they are scaled to match, they match each other very closely. There is a slight divergence lately since book per share has risen a bit faster than my valuation method, because earnings in the subs have been a bit weak in the last couple/few years. For these purposes it makes little difference.
The graph simply shows the real stock price (in today's dollars, log base 10), and the smoothe value. The value is scaled to give the average price:value ratio of 1 in the last 20 years.
This looks almost like a trend line, but there is a crucial difference: if the rate of growth of observable value changes, the smoothed line changes, over a maximum of 4 years but on average less than two. So there is no danger of extrapolating old fast growth rates.
Unsurprisingly, the ratio of price to estimated value is a pretty good predictor of whether you're about to see unusually good or unusually bad price changes.
In the last 20 years,
Cheapest 15% of the time: Two year forward real total return average 17.8%/year
Next 15% of the time: Two year forward real total return average 14.8%/year
Next 20% of the time: Two year forward real total return average 10.0%/year (just a little cheaper than average)
Next 20% of the time: Two year forward real total return average 8.7%/year (just a little more expensive than average)
Next 15% of the time: Two year forward real total return average 0.6%/year
Most expensive 15% of the time: Two year forward real total return average -5.4%/year
The current valuation level is at percentile 93.5, so around the middle of the "most expensive" bucket above.
The current ratio of price to the value line is 1.187, or 18.7% more richly valued on this metric than the 20 year average.
The 20-year-average valuation level today would correspond to $612733 per share ($408.49 per B). That price (or lower) might perhaps make a good re-entry target?? Of course that number will rise with share value and with inflation as time passes.
Those averages and returns are just observations about what happened in the past. The future may differ.
Jim
No. of Recommendations: 4
PS, this is an older image of the same valuation and smoothing method, from about two years ago.
It shows how similar the two valuation methods are, once they are scaled to match.
http://www.stonewellfunds.com/PriceAndWMAofValue.p...The single "value" line in the image in the first post is the simple average of the two methods.
Jim
No. of Recommendations: 1
Jim,
Nice work but just for clarification is your 16-quarter weighted moving average equally weighted?
No. of Recommendations: 3
Nice work but just for clarification is your 16-quarter weighted moving average equally weighted?
No, it's a WMA in the sense that chart people mean it. The most recent quarter is weight 16, the second most recent at weight 15, ramping down to weight 1 on the oldest quarter of the 16. So recent data counts the most, but doesn't totally dominate.
If you use equal weighting there is more time lag on average, and you can get meaningless discontinuities when an anomalous quarter shifts from being 16 to 17 quarters earlier.
A small note for those who like to be bullish: a bit of inflation as we saw recently increases the fair P/B ratio just a little. The reason is that still-productive fixed assets, depreciating or not, are rising in nominal output but not rising in nominal book value. The value of any contracts with revenues in fixed dollars falls, but I think Berkshire is smart enough not to have too many of those. The biggest exposure is probably any regulated rates which are not allowed to rise as fast as monetary inflation.
For those who like to be bearish, the big concern is the dire rate of increase of observable value among the operating subsidiaries. Both BHE and BNSF are doing poorly at the same time, and the problems may not be merely cyclical. Taking the whole group, rolling year net earnings on operating subsidiaries are up only 0.81%/year in the last 3 years, and under 4%/year in the last six years. Those figures include cyclical adjustment on underwriting profit so that volatility doesn't ruin the comparison.
Jim
No. of Recommendations: 2
The 20-year-average valuation level today would correspond to $612733 per share ($408.49 per B). That price (or lower) might perhaps make a good re-entry target?? Of course that number will rise with share value and with inflation as time passes.
I just wanted to add to that: the comment above is interesting and perhaps useful, but should not be confused with an estimate of the intrinsic value of a share.
Maybe Berkshire has been undervalued for most of the last 20 years, or overvalued. The comment is merely about the observed average valuation level, for better or worse.
Jim
No. of Recommendations: 1
"I just wanted to add to that: the comment above is interesting and perhaps useful, but should not be confused with an estimate of the intrinsic value of a share."
Nice analysis. Thank you for sharing, Jim.
I'm trying to infer your estimate of current intrinsic value. If I understand you correctly, your 20-year average valuation is intended to give a smoothed valuation level that tracks IV, but which is not necessarily equal to IV. It's apparently the average of your two and a half valuation and 1.5 times the 16 quarter WMA of BV. Is this correct?
Your current valuation level is $612,733, which corresponds to 1.554 times the 16 quarter WMA book value, which I calculate to be $394,284. It is also 1.326 times the March 31 BV/share.
From our recent discussion of IV/BV I understood, or misunderstood, your estimate of current IV to current BV to be flat over the last 25 years at a value of about 1.60. Certainly not 1.32 in Dec 1999 as I suggested.
Would you mind straightening me out? What would be your estimate of current IV to current BV?
FWIW, I'm still holding with an IV estimate of about 1.54x BV, based on a couple of models, but I would not argue with any estimate within 8 or 9 percent of that number.
Thanks for your input.
rrr12345
No. of Recommendations: 0
"FWIW, I'm still holding with an IV estimate of about 1.54x BV, based on a couple of models, but I would not argue with any estimate within 8 or 9 percent of that number."
Here are some other estimates:
person, model, IV estimate, date, IV/BV
Tilson, two-column, May '25, $743K, 1.61x BV
Greg Warren, Morningstar, Apr '25, sum of the parts, $730.5K, 1.58x BV
Bloomstrand, sum of the parts, Dec '24, $783K, 1.73x BV
Highest repurchase price, 1.57x BV
rrr12345, weighing machine, Mar '25, $710K, 1.54x BV
Would other folks here please add more estimates, whether their own or ones they're seen? It would make an insightful histogram. I used to keep track of posted and published IV estimates. Sometimes I'd find as many as 25 estimates, representing over a half dozen different models, including two-column, look through earnings, DCF, sum-of-the-parts, etc. Typically the estimates had a more or less normal distribution with a standard deviation equal to about 9% of the mean. Thanks.
No. of Recommendations: 1
I'm trying to infer your estimate of current intrinsic value. If I understand you correctly, your 20-year average valuation is intended to give a smoothed valuation level that tracks IV, but which is not necessarily equal to IV. It's apparently the average of your two and a half valuation and 1.5 times the 16 quarter WMA of BV. Is this correct?
Yes. I do two valuation methods. I scale each of them to get the best fit to observed 20 year price (inflation adusted). Then I take a simple average of the two scaled smoothings.
Your current valuation level is $612,733, which corresponds to 1.554 times the 16 quarter WMA book value, which I calculate to be $394,284. It is also 1.326 times the March 31 BV/share.
I'll take your word for it : )
I admit that I do another adjustment that I didn't mention in that particular post. The rationale is that when book drops during a recession, it's unlikely to be a lasting drop in value. Book per share just before a drop might be a pinch exuberant, as in 1999, but probably not wildly so. So during a drop in book per share I limit the drop in my "adjusted book" to be no more than 4% of the peak to date. Then I feed those figures into the WMA.
From our recent discussion of IV/BV I understood, or misunderstood, your estimate of current IV to current BV to be flat over the last 25 years at a value of about 1.60. Certainly not 1.32 in Dec 1999 as I suggested.
I think I was speaking of flat in the sense of "no net change", same at start and end. The middle has been very different. The average P/B in the last 20 years is just a hair under 1.40. (using peak-to-date known book, which is generally a better measure of value).
Would you mind straightening me out? What would be your estimate of current IV to current BV?
I can't really say. The only thing my calculations say is that the current price has a relationship to this trend line that is 19% higher than the average such relationship in the last 20 years.
Perhaps surprisingly, I don't usually try to calculate IV as such. As you note, my metric is something that correlates with IV, and rises at the same rate, but is some unknown multiple of true IV. That's because (a) it's impossible to come up with a multiple that's correct for true IV so everybody would pick a different one, and (b) a tracking metric is all I need for my two purposes: how fast is value growing, and what's the current relationship between price and value estimate compared to what is historically typical?
I could make a stab at true IV in the past. For me, intrinsic value is the price that you could pay and sell 5-10 years later at an "average" market multiple and make inflation + 6.5%/year, basically Siegel's constant. So I could tell you what my rough idea of IV was for any date more than 10 years ago : )
Take the average real price in the last 5 years scaled to 2015 dollars, divide by 1.065^7.5= 1.6037, and that's the price 10 years ago that (with hindsight) would have been true IV in my books.
Jim
No. of Recommendations: 1
Fun with crude numbers
"...the ultimate safe withdrawal rate is to liquidate no more than the amount the value of the shares has increased... You could withdraw money forever, or at least as long as the real value kept rising"
......
"...the ratio of price to estimated value is a pretty good predictor of whether you're about to see unusually good or unusually bad price changes.
In the last 20 years,
Cheapest 15% of the time: Two year forward real total return average 17.8%/year
Next 15% of the time: Two year forward real total return average 14.8%/year
Next 20% of the time: Two year forward real total return average 10.0%/year (just a little cheaper than average)
Next 20% of the time: Two year forward real total return average 8.7%/year (just a little more expensive than average)
Next 15% of the time: Two year forward real total return average 0.6%/year
Most expensive 15% of the time: Two year forward real total return average -5.4%/year"
So, I wondered: as the price:EV has been a pretty consistent long-term trend, then yet another way to think about the sensible "ultimate safe withdrawal rate" would be to add up the last 20 years of results:
0.15 times 0.178 equals 0.0267
0.15 times 0.148 equals 0.0222
0.2 times 0.1 equals 0.02
0.2 times 0.087 equals 0.0174
0.15 times 0.006 equals 0.0009
0.15 times -0.054 equals-0.0081
...with the sum of the RH column being 0.0791, implying an average 7.9% pre-inflation annual withdrawal would have been a sensible USWR 2005 - 2025
As the prices are not normalized for inflation, then the average CPI over the last twenty years of 2.5% needs to be subtracted, for a net USWR of 5 - 5.5%
Of course, this includes the highly questionable assumption that the next 20 years resemble the last 20, as regards Berkshire and the US economy in general and the CPI (which hasn't seen a 2.5% year since 2020).
But hey: you go to war with the data you have.
Anyhow, it further affirms my strategy of selling 1.125% current BRK holdings/quarter* being reasonable for at least the next few years, or until the effects of current policies begin to be reflected in market conditions.**
-- sutton
(*timing occasionally varies to current price/last known book: I sold the June 30 tranche in mid-March @ $522; conversely, I sold none at all during peak pandemic weirdness)
(**please, YWH, let this be at least a month or two before the midterms)
No. of Recommendations: 0
As an example, here is the list of IV estimates that I found for Dec 31, 2012. The document shows a histogram of the estimates as well as a table showing the sources. BV as of Dec 31, 2012 was $114,214, so the average IV estimate of $175K corresponded to 1.53x Dec 31 BV.
https://drive.google.com/open?id=1jR1lLKl3p8j0PnVU...
No. of Recommendations: 0
No. of Recommendations: 1
As an example, here is the list of IV estimates that I found for Dec 31, 2012.
Ahh, I remember the time well. Berkshire got cheaper and cheaper and cheaper, with the price having gone nowhere since 2017 but value steadily accumulating. I spent much of 2012 basically doubling my position, figuring that *eventually* the weighing machine would kick in. 2013 and 2014 were most satisfying.
Jim
