No. of Recommendations: 33
This is really just a follow-up to a previous post a few days ago, thinking about how much profit we should attribute to Berkshire's underwriting operations. Remember that, in his '5 groves approach', Buffett just ignores underwriting; many others assume it will be zero in an average year, or at least leave it at zero to be 'conservative', while some take a stab at what an average level is. Here's my stab.
I personally prefer to put in my best estimate for an average year's underwriting gains, rather than purposefully underestimating the number at zero. The other day, I looked at 15 years of underwriting earnings (pre-tax), just because beyond 15 years back, I got sick of getting the numbers from annual reports every 3 years, but I admit that 15 was pretty arbitrary, and I could reasonably be suspected of cherry-picking. So I decided I would go back to 2001, both because it includes a very bad year (2001) and because any prior data is really a qualitatively different Berkshire, before the huge transformative merger with General Re which closed in 2000.
So I have 22 years of data (2001 to 2022), and here's what I get:
Earned premiums and pre-tax underwriting earnings:
2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012
Premiums 74645 69478 63401 61078 57418 60597 45881 41294 41253 36684 34545
Earnings -98 936 838 417 2010 -3239 2131 1837 2668 3089 1625
% -0.1% 1.3% 1.3% 0.7% 3.5% -5.3% 4.6% 4.4% 6.5% 8.4% 4.7%
2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001
Premiums 32075 30749 27884 25525 31783 23964 21997 21085 21493 19182 17905
Earnings 248 2013 1460 2792 3374 3838 53 1551 1618 -400 -4067
% 0.8% 6.5% 5.2% 10.9% 10.6% 16.0% 0.2% 7.4% 0.0% 0.0% -22.7% 22-yr average: 3.21%
By the way, the 15-year average was pretty close: 3.6%, going back to 2008. 2007, 2006, and 2004 were pretty good, too, but 2001 was decidedly not.
So there you have it: my best estimate of pre-tax underwriting earnings is 3.2% pre-tax, in the post-Gen Re era. If we get a 9-11 size disaster every 20 years, we might expect to have that average, unless Berkshire's conservatism in underwriting policy-writing has changed (which I doubt.) I actually thought it would be lower.
On a technical note, I think it's better to apply the 3.2% average number to the current earned premiums ($74.6b, although this will jump higher with the Alleghany purchase), rather than calculate the average underwriting earnings, since these have gone up by more than 4 times over the last 22 years. 3.2% of $74.6b is $2.3b. I noticed this morning that Whitney Tilson uses $1.4b in pre-tax underwriting earnings for his estimate of current value; this would correspond to about 1.9% in pre-tax underwriting earnings, quite a bit lower than the 22-year average.
How important is this? Maybe not that important, but not completely insignificant, either. Buffett's estimate of 2022 'operating earnings'* is $30.8b:
The company's operating earnings ' our term for income calculated using Generally Accepted Accounting Principles ('GAAP'), exclusive of capital gains or losses from equity holdings ' set a record at $30.8 billion.This is after tax; since 2018, Berkshire's average effective corporate tax rate is 20%. $2.3b in underwriting earnings would be $1.9b after tax, which would have added 6.2% to the $30.8b in operating earnings. Ok, so I see that $2b is now almost a rounding error for Berkshire...
Regards, DTB
*By the way, looking at this more carefully than Saturday when I read the letter, I see that Buffett's definition of operating earnings is not what I thought it was: it is not the earnings of the operating companies, it is just something closer to old GAAP, with no unrealized capital gains in the income statement. But it does seem to include things like dividends from the equity portfolio. I just mention it in case anyone else has been making the same error I was making.
