Post by T Bhat on Nov 5, 2011 22:51:59 GMT -5
Berkshire intrinsic valuation from roberts in yahoo board:
1. Extrapolation of John Kish's estimates: John didn't publish an estimate for
Q2. I hope he's OK. As of Q1 John's "conservative" estimate was
$165.6K/A-share. BV is down 0.2% since Q1. If we adjust IV down by 0.2%, we
get $165.3K.
Probably a better estimate comes from regression of John's IV estimates over the
30-year period Dec 1981 to Dec 2010. Regression of John's estimates versus book
value (log-log) gives a very good fit, with r-squared = 0.9947. That is, the
single variable, book value, is sufficient to explain 99.47% of John's estimates
over a range from $1K/A-share in Dec 1981 to $165K/A-share in Dec 2010.
Extrapolation of the trendline to Q3's book value gives an IV of $160/A-share.
2. Stock price regression (price versus BV): "Over the long term the stock is
a weighing machine." Better than plotting stock price versus time is to plot
the stock price versus book value. (The appropriate plot is log-log.) The
trendline value of price versus BV depends on the period of time included in the
graph, but using the period from Dec 1981 to Sept 2011 gives an r-squared of
0.9803 and a trendline value at Q3's BV of $158K.
3. Investments plus 12x pre-tax, non-insurance operating earnings: Over the
period from Dec 1995 to present this "two-column" method has matched John Kish's
conservative estimates very closely, with no adjustments made to earnings and no
change in earnings multiple over the 17-year period. The correlation with
John's estimates is 1.00. As of Q3 the two-column model using 12x pre-tax,
non-insurance operating earnings (I multiplied the pre-tax operating earnings
for the first nine months of 2011 by 4/3 to get a four-quarter estimate.) gives
an IV estimate of $165K.
4. Bumping up the Dec 2010 estimates by the change in BV: The average of 28
posted or published IV estimates for Dec 31, 2010 was $150K. The standard
deviation was $21K. The median estimate was $152K. Since year-end BV has
increased by 1.49%. If we bump up the average estimate of $150K by the increase
in BV since year-end, we get an IV estimate as of Q3 of $152K.
My own estimate as of Dec 31 was $153K. The range using various models and
assumptions was $139K to $165K. If I bump up my own year-end estimate of $153K
by the increase in BV since year-end, I get an IV estimate as of Q3 is $155K.
The average IV estimate from the above methods is $160K.
1. Extrapolation of John Kish's estimates: John didn't publish an estimate for
Q2. I hope he's OK. As of Q1 John's "conservative" estimate was
$165.6K/A-share. BV is down 0.2% since Q1. If we adjust IV down by 0.2%, we
get $165.3K.
Probably a better estimate comes from regression of John's IV estimates over the
30-year period Dec 1981 to Dec 2010. Regression of John's estimates versus book
value (log-log) gives a very good fit, with r-squared = 0.9947. That is, the
single variable, book value, is sufficient to explain 99.47% of John's estimates
over a range from $1K/A-share in Dec 1981 to $165K/A-share in Dec 2010.
Extrapolation of the trendline to Q3's book value gives an IV of $160/A-share.
2. Stock price regression (price versus BV): "Over the long term the stock is
a weighing machine." Better than plotting stock price versus time is to plot
the stock price versus book value. (The appropriate plot is log-log.) The
trendline value of price versus BV depends on the period of time included in the
graph, but using the period from Dec 1981 to Sept 2011 gives an r-squared of
0.9803 and a trendline value at Q3's BV of $158K.
3. Investments plus 12x pre-tax, non-insurance operating earnings: Over the
period from Dec 1995 to present this "two-column" method has matched John Kish's
conservative estimates very closely, with no adjustments made to earnings and no
change in earnings multiple over the 17-year period. The correlation with
John's estimates is 1.00. As of Q3 the two-column model using 12x pre-tax,
non-insurance operating earnings (I multiplied the pre-tax operating earnings
for the first nine months of 2011 by 4/3 to get a four-quarter estimate.) gives
an IV estimate of $165K.
4. Bumping up the Dec 2010 estimates by the change in BV: The average of 28
posted or published IV estimates for Dec 31, 2010 was $150K. The standard
deviation was $21K. The median estimate was $152K. Since year-end BV has
increased by 1.49%. If we bump up the average estimate of $150K by the increase
in BV since year-end, we get an IV estimate as of Q3 of $152K.
My own estimate as of Dec 31 was $153K. The range using various models and
assumptions was $139K to $165K. If I bump up my own year-end estimate of $153K
by the increase in BV since year-end, I get an IV estimate as of Q3 is $155K.
The average IV estimate from the above methods is $160K.