Fairfax 2011 Annual Report & Chairman's Letter

[Grenville]

The annual report and chairman's letter are out:

 

Chairman's letter:

http://www.fairfax.ca/Theme/Fairfax/files/Letter%20to%20Shareholders%20from%20Annual%20Report%202011%20FINAL_v001_j939zn.pdf

 

2011 Annual Report:

http://www.fairfax.ca/Theme/Fairfax/files/Annual%20Report%202011%20Final_v001_j20lz3.pdf

---

 

Haven't dug in yet, should be good reading for the weekend.

 

[sdev]

Woohoo!

[beerbaron]

Was I asleep in the last year? When did they start Pacific Insurance (Malaysia)?

 

BeerBaron

[racemize]

here's the cumulative pdf:

 

http://dl.dropbox.com/u/14968/farifax%20(2012).pdf

[Parsad]

Another terrific letter!  Alot of good information.  No comments for you Fibrek shareholders.  You'll have to wait to the AGM to ask him anything.  Cheers!

[rijk]

wonderful letter.....

 

at the top of page 4, Odyssey Re has a CR of 97.5% for 2011 while in the table at the bottom the 2011 CR for Odyssey Re is 116.7%?

 

regards

rijk

[gordoffh]

As a large holder of SD, seeing Tom Ward mentioned in the letter was certainly nice to see.

[Grenville]

wonderful letter.....

 

at the top of page 4, Odyssey Re has a CR of 97.5% for 2011 while in the table at the bottom the 2011 CR for Odyssey Re is 116.7%?

 

regards

rijk

 

97.5 refers to their property line of business where the GWP is 798. The 116.7 refers to all lines of business and the NPW is 2089.7 for 2011.

[valuecfa]

A bit of market timing on Fairfax's part:

 

"In time, we will remove our equity hedges as the risks that we see get discounted in common stock prices. The major risks we see

are in the next three years, as we expect common stocks to do very well in the next decade."

 

Also, a slight bit contradictory to expect common stocks to do very well over the next decade while also seeing a fair chance of cumulative deflation over the next 10 years.

[twacowfca]

A bit of market timing on Fairfax's part:

 

"In time, we will remove our equity hedges as the risks that we see get discounted in common stock prices. The major risks we see

are in the next three years, as we expect common stocks to do very well in the next decade."

 

Also, a slight bit contradictory to expect common stocks to do very well over the next decade while also seeing a fair chance of cumulative deflation over the next 10 years.

 

Looks like they expect a big pullback in common stock prices in the next three years with potential for deflation and then a rebound with very good returns over ten years.

They appear to have timed their exit from long treasuries near perfectly as the potential for more gains going forward is slight.

 

Their runoff results are amazing!

 

They lost $1.5B on their crummy value trap and bottom fishing stocks in a flat market overall.  The equity hedges didn't help much.  And they still had a 6.9% investment return.  Unbelievable!

[rijk]

grenville, thanks for clarifying

 

regards

rijk

[Grenville]

Development that I thought was notable:

 

Sam Chan - became president of Alltrust in China at Chairman Henry Du's request and Scott Donovan, retired CFO of ORH, will also support Sam

 

---

 

Enjoyed the letter, the EU situation (~25% of world GDP) and China (~10% of world GDP) worry me. The lack of levers really makes me hope the US consumer pulls us through the next few years.

 

The level of detail and the references to individuals and the color on the way things developed is awesome. It's a unique view to have as a non employee shareholder. The international expansion will give us a nice view on the developments around the world.

 

 

[link01]

Looks like they expect a big pullback in common stock prices in the next three years with potential for deflation and then a rebound with very good returns over ten years.

 

even interpreted this way it still seems like they are timing eventual stock market outcomes. maybe not with the 10 yr outlook, but certainly with the 3 yr where they seem to suggest stocks will discount the deflation risks during that timeframe, setting up the potential for attractive risk adjusted, mean reverting returns after.

 

this surprises me somewhat considering that the case-shiller model still pegs the broad market indices to be 35-45% overvalued based on long term real trend earnings. so, is this another way of saying watsa expects a 35% or more swoon in stocks prices within 3 yrs? or perhaps that he parts company with shiller's valuation model here? or, maybe, like buffett, he sees a major adrenalin shot to the economy & to animal spirits from a rebounding housing sector at long last!

 

here some interesting comments i saw recently concerning real trend earnings:

 

<<There are many, many different ways to go about this, but the central

estimate seems to be "fair value" of about 933 for the S&P for end Feb 2012,

actually measured in December 2011 dollars.

For "fair value" I simply use the observed historical average multiple of trend E,

since the multiple appears to be mean reverting to a typical central range.

So, 933 is the valuation level that would make the current multiple of

my estimate of today's on-trend sustainable earnings equal to the

average observed multiple of that trend line since 1940.

The historical average multiple isn't particularly in doubt, it's 13.58x.

The problem is trying to estimate today's level of trend sustainable earnings.

I come up with the figure of about $68.75-69.50 now.

(actually trailing year reported earnings are about $87, running above trend).

 

Flipped around, for today's market level of 1363 to be fair value, the on-trend

trend real earnings would have to be spot on $100 right now in Dec 2011 dollars.

That won't be the case for many, many years.

 

A very brief recap of the process:

Start with a history of all US earnings, trailing 12 months, using the

monthly data series from Mr Shiller's web site since 1871.

Convert all of those to constant dollars with a CPI adjustment.

Smooth the series by replacing each month's value with the average

for many years before and after that. That gives you the perfect

on-trend earnings figures for the past, though not for lately.

Calculate the historical average ratio of price to then-current on-trend earnings.

(actually I average the trend earnings yield which is more numerically

stable to get 7.363%, then invert it to get the P/E ratio of 13.58x).

Build a number of different models based on only the data known at the

time to try to estimate the current on-trend level.

Test all those models against the now known perfect values to see which

models worked the best on average in the past.

Build a consensus model with a mix of about 5 of the best models.

Run those models for today to see what they say trend earnings are likely to be.

 

I have done the whole process separately with a different data source,

and it comes to about the same conclusion.

 

The final result of fair value 933 depends on whether you look at

the historical average multiple or the historical median, and whether

you look at the entire data series since 1871 or some more recent subset.

These different approaches give figures ranging from 900 to 1050,

but 933 seems to be the most sensible estimate I have right now.

 

The S&P is not 47% overvalued.

 

You might be right, I might be wrong. I sure hope so, as I'm quite fully invested (though hedged too).

But lay out your evidence. Where's the logic hole in what I did?

If you come up with a better way, I'm all ears and will gladly switch!

For example, GDP-to-index-level is by comparison an extremely crude yardstick

for a half dozen reasons. It's cute and easy, but it's dead easy to do much better.

 

Does the fundamental value of a collection of stocks come from anything

other than their net after tax trend real earnings? In a word, no.

 

Should we expect the market to trade at a multiple of earnings that is

far from the historical norm? Why would that be?

It has been mean reverting for a very long time, though only very slowly.

The multiples might stay higher than average for 20 years, or might not,

but if one is estimating central expected long run returns from an

initial purchase time and price (the purpose of my work), then that won't help.

The most sane central expectation is that the average multiple in the

next 100 years will probably be a lot like the average multiple in the last 100.

 

Is the trend real earnings line a whole lot higher than I estimate it?

Sure, maybe, try it yourself and graph it. Figure out the likely trend line yourself.

But the central trend line of a squiggly data series won't go through a

new all time high such as we just saw no matter how you torture it.

So, current earnings are above trend, it's only a matter of how much.

 

Holes in the data source? Sure, CPI is a common example. But I'm

looking at ratios of inflation-adjusted earnings to inflation-adjusted

prices, so even if the CPI is off by a mile it mostly cancels out.

 

The best approach that I can think of for defending a higher valuation

is that the weak hand of global labour versus the strong hand of

global capital in this 20-30 year period is going to stay the way it

is for a long time, meaning the largest global companies are going

to get a disproportionate share of the GWP pie for a long time to come,

so we might see an above-trend earnings decade or two as happened after the war.

Plus, since a lot of those firms are in the S&P 500 you'll see a

continuation of the trend of earnings growing faster than US GDP.

That still wouldn't make today's market level a typical market valuation

level, but it would mean that it's only slightly overvalued.

 

By the way, nothing about this has anything useful to say about the

likely direction of stock markets in the short term, say under 3 years.

It's not a sell signal on the S&P for a short term trader.

But it's an excellent guide for what to expect in the next 5-10+ years.

Based on the typical results in the past starting from levels of

price-to-trend-earnings similar to today's, the likely expectation

for the next 10 years is about 0-1%/year after dividends and inflation.

That's with 90th percentile outcome +5.6%/year, 10th percentile outcome -2.2%/year.

Since high valuations tend to last for a while, the short term looks a little better:

3 years out central expectation is about 3%/year after dividends and inflation,

but the range of outcomes is so wide that it's not very predictive:

90th percentile +21.1%/year and 10th percentile -8.1%/year.

 

Jim (aka, mungofitch at the TMFbrk fool board)>>

[returnonmycapital]

At the bottom of the first page of the annual letter, Prem writes:

 

"The $1 billion in catastrophe claims in 2011 cost us 19.3 percentage points on our combined ratio versus an anticipated cost of approximately six percentage points in an average year."

 

They're combined ratio for 2011 was about 120% (if you use the consolidated income statement). Does that mean they anticipate a combined ratio of 106% in an average year?

 

If so, that would imply an anticipated yearly average cost of float of approximately 2%.

 

 

[StubbleJumper]

At the bottom of the first page of the annual letter, Prem writes:

 

"The $1 billion in catastrophe claims in 2011 cost us 19.3 percentage points on our combined ratio versus an anticipated cost of approximately six percentage points in an average year."

 

They're combined ratio for 2011 was about 120% (if you use the consolidated income statement). Does that mean they anticipate a combined ratio of 106% in an average year?

 

If so, that would imply an anticipated yearly average cost of float of approximately 2%.

 

Nope.  Prem meant that cats normally should account for 6 points of CR, full stop.

 

In an average year, Prem is probably hoping to write a 98 which would include the 6 points.  I'd say averaging 102 including the 6 points would be more realistic.

 

SJ