Prem's Comments At ACG Toronto Today

[niels12think]

1) Performance.  The BVPS of FFH has gone similarly to BRK but 1) in a much more lumpy way (FFH was in no great shape some years back and things could have taken a turn for the worse fairly easily)  2) on a much smaller duration  and 3) with much more leverage (compare the equity/assets ratio even now of BRK vs FFH).

So yes, the returns for Fairfax is more lumpy - but not that much.

In fact, the lumpyness in Fairfax' return on book might be thought of as additional return when compared with Berkshire.

 

Actually, we can't really know that the returns of Fairfax are more lumpy than Berkshires - it just looks that way.

But it may be due to chance.

Or it may be due to some bad aquisitions - which may not be repeated going forward; for the simple reason that the persons involved may have learned from the lessons.

 

Cheers

 

[oldye]

Hedging isn't is anything free...Berkshire is rich enough to accept the volatility and profit from it, if anything Fairfax's success predisposes them to hedging and I'm sure they'll be extra careful.  I

[frog03]

Actually, if we look at the last almost 24 years, then the book value per common share of Fairfax has increased 26% annually (taking dividends into account). For comparison, Berkshire has increased book per share by 20% annually over the 25 year period from 1984-2008.

As for the lumpyness, Berkshires change in book value from year to year have a Gini coefficient of 0.42 from 1984-2008, where the Gini Coefficient for Fairfax' book increase is 0.50 for the period 1985-2009´Q3.

So yes, the returns for Fairfax is more lumpy - but not that much.

In fact, the lumpyness in Fairfax' return on book might be thought of as additional return when compared with Berkshire.

 

24 years is pretty long duration - but for obvious reasons it is smaller than Berkshire' full history.

 

>>>  Actually, you probably would want to compare the first 24 years of BRK vs the first 24 years of FFH.  Otherwise, FFH has a huge size advantage (being so much smaller makes it easier to get good returns).

[niels12think]

Actually, if we look at the last almost 24 years, then the book value per common share of Fairfax has increased 26% annually (taking dividends into account). For comparison, Berkshire has increased book per share by 20% annually over the 25 year period from 1984-2008.

As for the lumpyness, Berkshires change in book value from year to year have a Gini coefficient of 0.42 from 1984-2008, where the Gini Coefficient for Fairfax' book increase is 0.50 for the period 1985-2009´Q3.

So yes, the returns for Fairfax is more lumpy - but not that much.

In fact, the lumpyness in Fairfax' return on book might be thought of as additional return when compared with Berkshire.

 

24 years is pretty long duration - but for obvious reasons it is smaller than Berkshire' full history.

 

>>>  Actually, you probably would want to compare the first 24 years of BRK vs the first 24 years of FFH.  Otherwise, FFH has a huge size advantage (being so much smaller makes it easier to get good returns).

 

I was comparing the same period in order to factor out the different economical environments of the periods.

If we look at the first 24 years of Berkshire book value per share growth, then we have annualized returns of 23.3%, distributed as follows

Year	%-change
1965	23.8
1966	20.3
1967	11
1968	19
1969	16.2
1970	12
1971	16.4
1972	21.7
1973	4.7
1974	5.5
1975	21.9
1976	59.3
1977	31.9
1978	24
1979	35.7
1980	19.3
1981	31.4
1982	40
1983	32.3
1984	13.6
1985	48.2
1986	26.1
1987	19.5
1988	20.1

source: page 4 in the 2008 annual report; link: http://www.berkshirehathaway.com/2008ar/2008ar.pdf

 

Ie. Berkshires returns during their first 24 years were almost 3 percentage points a year lower than what Fairfax produced.

 

During both Berkshires first 24 years and during Fairfax's first 24 years, the S&P500 returned approx. 9% per year including dividends, but before taxes.

For both Berkshire and Fairfax, the results are after corporate tax.

 

However, it is true that during the first 24 years of Berkshire, the lumpyness of the results were lower: Gini coefficient of 0.37.

But please keep in mind, that Berkshires results are calculated based on restated numbers (similar accounting principles throughout the period), whereas Fairfax' results have some artificial lumpyness because there have been changes in accounting principles and changes in the base currency - both of which adds to the apparent lumpyness, but not to the intrinsic lumpyness.

 

Cheers :)

[Uccmal]

To be fair you need to remove the first year of FFHs operation (180%) and call it a fluke.

 

Also if you smoothed it for the effects of financial engineering you will probably find that FFH and BRK have identical records of return on equity.  To clarify this and to Frog03s point FFH has issued shares at 3 x book and bought back shares at below book, neither of which Buffett has ever done (at least not in those extremes).  I dont think this is an ethical or integrity issue - just value investing.  If Prem can generate institutional interest at such huge p/b then go for it, I wont be buying. 

 

I am not going to work it out because it is a waste of time.  Suffice to say FFH and Berk continue to be very good investments.  I personally think FFH has greater, or perhaps faster, upside in the near term, while Berk will periodically get caught by the large numbers rule. 

[Crip1]

Guys, this is becoming a Bobby Orr vs. Wayne Gretzky discussion.

 

We can compare and contrast the relitive merits of Buffett and Watsa (or Orr and Gretzky for that matter) ad infinitum, but the key question is whether or not you want them on your "team".

 

My answer for Buffett and Watsa is "Yes".

 

-Crip

[niels12think]

To be fair you need to remove the first year of FFHs operation (180%) and call it a fluke.

 

If you replace the best result (the first year) with an average result of the rest, then the extra lumpyness of Fairfax cancels out and the Gini coefficient is reduced to 0.35, i.e. meaning that Fairfax have had the same (or slightly lower) lumpyness in earnings as Bershire during the first years.

 

Also, if you back out the best result of Fairfax, but doesn't do the same with Berkshire, then the annualized returns remain comparable: Fairfax 22.8% and Berkshire 23.3%.

 

Both results are exceptionally good in my opinion - and to answer Crip1' question: Yes - I would want both Buffet and Watsa on my team :)

 

Cheers !

 

 

[shalab]

Can you please explain how did you calculate your numbers?

 

cheers!

 

[niels12think]

Can you please explain how did you calculate your numbers?

 

Nothing fancy:

For Berkshire, the annual change in book value per share is available on page 4 in the 2008 annual report. link: http://www.berkshirehathaway.com/2008ar/2008ar.pdf

 

For Fairfax, the different book values per common share are available in the annual reports. From these, I add in the dividend paid per common share and calculate the change.

I used the numbers in CAD for the first period and then switched into USD. Restated numbers where used where available.

The 2009 numbers includes only the first nine months and is calculated based on the actual number of shares outstanding at the end of Q3 - not the average number of shares effectively outstanding.

All the different annual reports and the 2009Q3 report are available here: http://www.fairfax.ca/financial.htm

I get the following numbers for the annual percentage change in book value per share including dividends (Fairfax):

1986 183.17

1987 41.26

1988 21.75

1989 22.51

1990 39.32

1991 23.83

1992 10.98

1993 47.85

1994 24.59

1995 21.73

1996 63.38

1997 44.03

1998 47.18

1999 25.71

2000 0.84

2001 -18.17

2002 13.09

2003 29.13

2004 -7.77

2005 -14.66

2006 10.23

2007 54.11

2008 22.18

2009 37.24

 

In computing the Gini coefficents, the level is moved up if necessary (to eliminate negative numbers which would give misleading results) - and I used 24 and 25 buckets, respectively. Neither "smoothing" nor interpolation was used.

I know we would normally like to have a lot more measurements and that Ginis are not normally used in these situations, but I'm of the opinion that in spite of these things, the calculated Gini coefficents make sense,

although smaller differences may be due to chance and not really indicative of any difference in the real "lumpyness".

Gini coefficents - including formulas - are explained in great detail here: http://en.wikipedia.org/wiki/Gini_coefficient

 

In computing the annualized return, I calculated as though dividends could have been re-deployed without taxes and at the same rate of return as the remaining book value.

I know this is generally unrealistic as there would generally be taxes to pay on dividends and generally it's not possible to redeploy dividends at book value.

However, in the case of Fairfax, it has - and still is - possible to redeploy money at or below book.

The first nine months of 2009 was included in the annualized return calculation as 0.75 year.

 

Comments and questions are welcome as always :-)

 

Cheers

 

[niels12think]

Can you please explain how did you calculate your numbers?

In computing the Gini coefficents...

 

also, the Ginis were indexed to make them immediately useful for comparison.

 

Cheers

 

[niels12think]

Can you please explain how did you calculate your numbers?

 

sorry, I can see that I had an error in my calculation of the Ginis [i used A/B instead of A/(A+B)].

I now get the corrected values for the Ginis of the first 24 years as

BRK  0.23

FFH  0.35

FFH (excluding highest return) 0.27

 

therefore, BRK had somewhat lower lumpyness in book value growth than FFH.

 

thanks shalab, for pointing me to this  :)

 

cheers