No. of Recommendations: 17
I thought he laid it all out quite well in the 2018 letter:
https://berkshirehathaway.com/2018ar/2018ar.pdf
I follow what he wrote, use a multiple of earnings for the subs, include stocks and fixed income investments at market value, subtract some cash that will never be invested.
After discussions here I did add in a % of float for underwriting earnings (1.4% - the 10 year average).[1] It doesn't make much difference.[2]
That approach certainly sounds reasonable, but there's a lot of filling in the blanks from what Buffett wrote. For instance, unless you do some adjustments, evaluating the first grove non-insurance subsidiaries just based on the previous year's earnings seems too volatile.
The second grove, public equities, is straightforward for the equities, especially now that Apple is down to a size similar to the others and doesn't need any adjustment.
The third grove, joint ventures, is getting less and less important, as KraftHeinz dwindles into oblivion, Pilot is now (unfortunately) a fully owned business, and Berkadia seems to have become tiny (in the 2024 annual report, it was lumped in with 'Other', with a carrying value of $452m.) Actually, no, scratch that, as I wrote that, I thought, "But it's not so insignificant to Jefferies, a much smaller company who owns the other half", thought I. And indeed, the Jefferies annual report that says it earned $192m pre-tax in 2024 which obviously gives us the number for Berkshire, too. And for 2025, a reasonable adjustment might be to count their 28% share of KraftHeinz's approximately $2.6b in earnings but not to count the huge writeoff in Q2 which will make earnings temporarily negative.)
The fourth grove, the fixed income investments, is harder. Much of it corresponds to float, and Buffett didn't really say whether it should be included (despite the fact that it is counterbalanced by obligations) or not. But I think it is a pretty much permanent source of fixed income earnings, and I'm not so sure you even need to subtract any cash - is it really true that it will never be invested? Isn't it ALL invested, mostly in treasuries? I don't think there's any cash sitting in a chequing account earning 0% interest.
The fifth grove, I agree with you, should be counted, although in 2018, Buffett seemed to waffle:
"much of our ownership of the first four groves is financed by funds generated from Berkshire’s fifth grove– a collection of exceptional insurance companies. We call those funds “float,” a source of financing that we expect to be cost-free– or maybe even better than that– over time. We will explain the characteristics of float later in this letter."So he doesn't really say anything about underwriting, which clearly has been a huge contribution to Berkshire's success for decades.
So now that I've thought about this, here's my estimate for end of 2024, which we will be able to update next week when we have 2025 results. My preferred approach is to subtract out the things we can value fairly precisely (equities and fixed income, or groves 2 and 4), and then calculate the earnings from groves 1,3 and 5, using the average combined ratio for the last 10 years applied to current net premiums earned for grove 5. That gets me:
Grove 1: $24.271b net earnings of BNSF ($5.031), BHE ($3.730b) and MSR (Manufacturing, service and retail; $13.072b) p. 5)
Grove 2: $271.588b market value (p. K-102)
Grove 3: $1.519b net earnings of associates (p. 5)
Grove 4: $349.565b (p. K-66)
Grove 5: Net premiums were $88.257 in 2024, average underwriting margin in the last 20 years has been 3.46%[3], so I'm saying $3.058b in normalized underwriting earnings.
So now I subtract the equities and ALL the fixed income stuff (including cash) from the market cap, so $1068b minus $621b from groves 2 and 4, and get $447b for the groves that produce earnings. (Of course, groves 2 and 4 produce earnings too, but no double counting, so they have just been valued at market price, without looking at their earnings.) Those $447b in assets produced a total of $28.848b in earnings, which is a P/E ratio of 15.5. (The alternative method, giving some arbitrary multiple to these earnings, say 15, and then asking whether the stock is overvalued or undervalued, gives you a similar answer, but I prefer to just calculate what the market is implicitly paying for groves 1, 3 and 5, and then anyone can be the judge of whether that multiple is too high or not.)
Bottom line, if you think Berkshire's equity and fixed income portfolio is fairly valued, and if you think that 15.5 is a reasonable multiple for a bunch of earnings streams (largely BNSF, BHE, MSR, a little bit of KHC and Berkadia and Occidental, and $3b of underwriting earnings at the average underwriting margin), then Berkshire is reasonably priced.
I have no position right now. I have been keeping a fairly close eye on the company, as it was my biggest investment for 20 years until a few years ago. I intend to wait to hear from Mr Abel who has seemed much more impressive to the people who know him well than to me, so far. I am keeping an open mind, but to reinvest in Berkshire I would want to hear that Abel has a vision for repurchasing shares at a rate that is an order of magnitude higher than what Buffett seems to have had in mind. I share some of the concern that Abel is not primarily an investor, and that is job #1 for the Berkshire CEO, but I presume Buffett and Munger thought he would be up to this part of the job, or they wouldn't have chosen him. I don't like a lot of the choices Buffett has made in the last 10 years, so, different from some investors on this forum, Abel is the reason I
might buy shares again, not a reason to avoid them. And maybe if he sounds like he is not up to the job (public speaking does not seem to be his forte), Mr Market may give me a hand and allow me to buy those shares at a lower multiple.
Regards, DTB
Notes
(1) "After discussions here I did add in a % of float for underwriting earnings (1.4% - the 10 year average)." - I get 2.42% for the last 10 years, and 3.46% for the last 20 years. I just use insurance premiums earned (disclosed on p.K-68 last year) and divided by net underwriting gain (on p. K-33 last year). Maybe we are doing this differently? Or if you did the calculation 2 years ago, by my calculation it would have been 1.70%, much closer to your figure.
(2) I think it makes enough of a difference to be worth doing, at a 3.46% margin - they were 11% of all net earnings in 2024. Excluding those $3.058b in UW earnings would give us a P/E of 17.3 instead of 15.5.
(3) This is the fastidious part of the calculation, finding net premiums earned and net underwriting gains for the last 20 years. Of course, there has been no 9/11 in the last 20 years, so even 20 years may not be enough to be sure, especially for reinsurance. Reinsurance is now a much smaller part of Berkshire's overall business (shhrinking from 70% of insurance premiums in 2000 to 32% in 2024), and insurance itself has diminished from 57% of Berkshire's revenues to 23% in 2024, so while this was a major concern in the years after the GenRe acquisition, it's a small fraction of the risk now.
