Recommendations: 33
Here's a fun chart, based on a valuation model much like that of Whitney Tilson. To wit, a multiple of investments per share plus a multiple of pretax operating income per share which includes an estimate of underwriting profit. Unlike Mr Tilson, my analysis uses the same multiples every year. More signficantly, I use much smaller multipliers.
Here's my chart: stonewellfunds.com/TilsonValue.gif
The valuation model used in this chart is: 80% of investments per share (to account for the perpetual drag of cash), plus 9 times pretax earnings per share (long run average valuation for the average quality listed US company). These are very conservative multiples.
The operating earnings per share are adjusted in several ways. If Mr Buffett produced a twocolumn commentary in a given year, I started with that figure. For other years, I started with the pretax earnings with the following adjustments:  Remove otherthantemporaryimpairments, since they merely move earnings from the income statement to the equity lines  Remove derivative gains/losses, on the assumption that it will average out to breakeven over time  Remove investment gains/losses, as they are covered in the investments per share side.  Remove underwriting gain/loss as it needs cyclical adjustment.
Starting from either of those two numbers, I added in a figure for cyclically adjusted pretax underwriting gain/loss estimated as 0.8% cost of float. So, the general earnings model is "column 2 plus cyclical underwriting profit". Is 0.8% reasonable? Probably so. The average cost of float since and including a horrible 19992001 for Gen Re cleanup has been 0.57%; but 20032011 it has been 3.48%.
I made a couple of rather arbitrary adjustments for simplification, but nothing that changes the conclusions. For example, the pretax earnings for 20012004 are simply interpolations from 2000 and 2005. Shoot me.
The final figures are for the four quarters Q4/2011 through Q3/2012 inclusive. The red dot is yesterday's closing price of $130890.
One can see from the chart a few fairly obvious things.  This is a firm that has increased in value pretty steadily using a numerically stable estimate such as this, and continues to do so.  The slowing in value growth rate since the last 1990s is not nearly as big as one might guess from the slowing in share price growth. The rate of value increase since 1997 is 11.2%/year, which is 12.2%/year in the first half of that 14.75 year period and 10.4%/year in the second half. For comparison, SPY with dividends was 4.6% in the first half and 4.1% in the second half, so the advantage was 7.5%/year in the first half and 6.3%/year in the second half.  The current price is very cheap compared to this very conservative value estimate. As of the Q3 statements, the valuation on this model is $158,487.  Historically the market price has fallen below this line only very rarely, yet has been below it continually for the last two years.
FWIW, a pretty good model in the past for average price in a given year is around 1.116 times the figure this gave at the end of the prior FY. That would suggest the 12 months Oct 1 2012 through Sept 30 2013 "should" have an average price of $176870 at historical average valuation levels. This model zaps out the price bubble 19941998 via the simple expedient of modelling the minimum price in those years rather than the average. This makes the model more conservative by removing the influence of the biggest price bubble. Even so, we might not see those average multiples in future, but it's hard to argue that the price shouldn't achieve at least this very conservative valuation estimate line.
Extrapolation suggests that this value model (the blue line) will give a figure of ~$176k at end 2013, leading to an average price guess of ~$196k in 2014. The value model is quite conservative, the price figure less so.
Jim



Announcements
