Assessing Buffett's Recent Performance Using His Own Measure
Average Annual Gain for Period from 1965 to 2003:
Berkshire book value per share: 22.2%
S&P 500 including dividends: 10.4%
Buffett's record is particularly impressive when stated another way: $1,000 invested with Buffett in 1965 at book value would have grown to $2.5 million by the end of 2003, versus just $47,000 if invested in the S&P 500. Over this 39-year period, the end results differ by a factor of a little more than 50! This long-term track record of outperforming the market is what makes Buffett the renowned superstar investor of the 20th century.
Accolades aside, however, one aspect of Buffett's comparison that has always bothered me is that changes in book value are being compared to changes in market value. In other words, it looks like an "apples to oranges" comparison. Now, over a long enough period of time, book value and market value for a particular asset will tend to track one another, so that one can argue that the comparison is a fair one. However, over a shorter time period, wide swings in market value (typically resulting from investors' often-changing perception of the future) tend to make the comparison less meaningful.
In order to judge Buffett's performance over a shorter time horizon in an apples-to-apples framework, we can refer to book value per share data available from moneycentral.msn.com. Using historical book value figures for the 30 components of the DJIA to create a comparative benchmark (working with all 500 stocks in the S&P 500 would be too unwieldy an exercise for me to undertake!), I find:
Annual % Change in Book Value Per Share for the 8-year period 1996-2003:
Berkshire: 17.0% (using figures from Berkshire annual report)
DJIA: 11.1% (based on historical book value per share data)
However, this is still not apples-to-apples. Since payment of dividends reduces book value, we must add the dividends back into the calculation for the DJIA to get figures to compare with Buffett's:
DJIA Annual % Change in Book Value Per Share
+ Dividends as % of Book Value
= 11.1% + 8% to 9%
= 19% to 20%
(Note: For the DJIA, the average price-to-book ratio is currently 3.8, which implies a normal dividend rate of about 2.3%, corresponding to the dividend-to-book-value rate of 8% to 9% shown above.)
No similar correction for dividends is needed for Berkshire because Berkshire does not pay dividends.
The fair apples-to-apples comparison, then, is that the DJIA actually outperformed Berkshire by 2 to 3 percentage points (i.e., 19% to 20% for the DJIA vs. 17% for Berkshire) for the period 1996-2003, when compared using changes in book value per share. In other words, when assessed using his own measure, Buffett has been slightly underperforming the market in recent years.
Admittedly, the period 1996-2003 includes Buffett's worst and only year (2001) when the change in Berkshire's book value per share was negative (-6.2%) due to insurance reserve-related losses from General Re, the reinsurance company Berkshire had acquired in 1998. Given Buffett's otherwise consistent long-term record, it is most likely the case that 2001 was an anomaly, and that the remaining 38 out of 39 years in the period from 1965 to 2003 are much more representative of the true capabilities of the Oracle of Omaha.
Since long-term trends in a company's book value should ultimately drive market value, we should also look at market value performance over our 8-year study period:
Market Returns for 8-Year Period 1996-2003:
S&P 500: 7.5%
We see that, although Berkshire's book value may have lagged the DJIA in recent years when measured by Buffett's own "weighing machine," Berkshire's market value outperformed the DJIA by a wide margin. The market's "voting machine" apparently reveals the high expectations shareholders have on Buffett's uncanny ability to continue to beat the averages.
(Disclosure: I have owned Berkshire Hathaway shares since 1996 and consider this position one of my long-term holdings.)