Subject: Operating earnings
Mr Buffett tells us on a regular basis that we should be paying attention to operating earnings. I haven't, to date.

And in a sense I still haven't: Berkshire's definition of operating earnings is after-tax earnings, minus realized investment gains/losses and minus one-time impairments.
I used those figures, but did one adjustment which I really think is needed to get an idea of business progress: I replaced the wildly volatile actual underwriting result for each year with a cyclically adjusted estimate (the average of two estimates: % of float and % of premiums earned) which adds up to about the same long run profitability from underwriting.

So, here is a quick look.
Operating figures (with only my one underwriting adjustment) adjusted for inflation (today's money CPI=314.2), each prior figure taking the BLS' published average annual CPI figure for that year, not Dec 31.

If you take the logs of these figures and create a linear trend line, and exponentiate the result, you can estimate what the "on trend" figure might have been, giving rise to the comments below.

Figures in millions, total, constant dollars.
  2003     8,050 spot on trend
  2004     7,402
  2005     8,693
  2006    11,345
  2007    12,049
  2008    12,174
  2009    10,493 below trend by 9.5%, credit crunch
  2010    14,960
  2011    15,908
  2012    16,709
  2013    18,669
  2014    20,744
  2015    22,517
  2016    22,331 spot on trend
  2017    22,819
  2018    30,603
  2019    30,679
  2020    27,500 below trend by 9.5%, pandemic
  2021    32,725 spot on trend
  2022    34,915 only 1.6% below trend, maybe I have been complaining too much
  2023    34,729 below trend by 9.4%, wildfires?
  TTM     36,263 to Q2, still 9% below trend, but it's an overlapping stretch
The slope of the trend line is growth in operating earnings of inflation + 8.08%/year in this stretch.

**BUT**
All those figures above are the totals for the company. The per share figures are of more interest, since some capital allocation has gone to buybacks rather than to increasing top level operating earnings.

Here is the same table again.
Figure in dollars per share, real dollars.

  2003     5,237
  2004     4,810
  2005     5,641
  2006     7,353
  2007     7,783
  2008     7,859
  2009     6,762 below trend, credit crunch
  2010     9,077
  2011     9,637
  2012    10,170
  2013    11,356
  2014    12,627
  2015    13,702
  2016    13,581
  2017    13,873
  2018    18,650
  2019    18,880
  2020    17,811 below trend, pandemic
  2021    22,150
  2022    23,919
  2023    24,092 only 3.0% below trend, barely worth mentioning
  YTD     25,241

The slope of this is inflation + 8.276%/year, and it is a slightly better fit (higher r2). Things are doing pretty well.

It's also a slightly higher rate of value growth than my usual estimate of growth of real share value, which has usually been around inflation + 8.1%. It's a bit tricky to visualize the reason for the difference: this operating earnings view both does and does not include the investments per share side of the business: it includes dividend and investment income from the investments, but never includes any benefit from the rise in the value of shares. Basically it's an "everything but buy-low-sell-high" view of things. Consequently "operating earnings" might be an atypically poor view of the progress in 2024 because of the huge realized gain on Apple which was counted neither as an unrealized gain as it happened, nor as a realized gain in this period.

Jim