Fairfax 2026

[SafetyinNumbers]

2 minutes ago, Marco Van Basten said:

I think that if you were to mark to market their investments, the returns would be much higher...

These latent gains and the accounting rules that explain them are good reasons why equity returns might be above average when measured over 3-year periods. 

[Hamburg Investor]

25 minutes ago, dartmonkey said:

 

 

I think 9.3% returns on total iinvestments is probably too optimistic - the average for the last 6 years is 6.1%. The last 6 years show how variable this is, from year to year, compared to bond returns:

 

I calculate the last line by backing out the fixed income return from the overall return, and calculating the return on equity investments (which includes associates and consolidated non-insurance companies.) As you can see, the fixed income return was low in 2020-2022 and has recovered to about 5% now. But the equity investments are, as you would expect, all over the place. For instance, in 2021, returns were great, because they booked a huge gain in Digit Insurance. 2022 was terrible because it was a terrible year for stocks, with the S&P down 19%. And so on.

 

So I think extrapolating last year's great 9.3% return, and particularly its 18.8% equity investment component, is too optimistic.

 

But even so, with the leverage in place, low corporate costs and low interest costs, and great investments like Eurobank and Fairfax India trading at low valuations, I am pretty confident in their prospects. I think we already have

 

My model is still embryonic, but if I assign 13.2% to equity returns (the 6-year average) instead of 18.8%, with fixed income returns unchanged at 5.0%, we should still have overall returns of 7.6%,  a bit better than the 6-year average of 6.1%. And with already 2 major investment successes for 2026 (Eurolife - $350m, Poseidon - $866m, total of $1216m), after only 2 1/2 months we are well on our way to repeat something similar to last year's $3151m in total investment gains (9.3%) than the 6-year average. 

Thank you; I really like your table! It seems logic and  rather uncomplicated way to understand things. I've got some questions to you:
- I go with you, that 18.8% is nothing I would expect on average to be reachable. More like 12% or maybe a bit more over time on average. Maybe the 13% is a good number to start with. So ROE 13% on the equity part.
- At the same time hidden assets are huge; and while it's good for intrinsic value, that they are not in the books yet (as this means, that they haven't been taxed yet - so the tax deferral gives an extra - hidden - cost free leverage to FFH), at the same time Eurolife, Poseidon... are only a small part of hidden equity returns of the past years, that now come to light. The value creation of Eurolife, Poseidon etc. hasn't happened overnight but has been growing steadily and all the time. The value has been created, but we just haven't seen it in the books. So we don't find that in the numbers and tables (with the rare exception of the $3bn+ of hidden value in the stocks, that are only missing for accounting reasons; but even those 3bn are not calculated in official returns, earnings, ROEs). This $3bn (and I guess more than that of real hidden value) is missing everywhere in the numbers - and so in your table.
- So for that alone you could adjust the equity investments of the last 6 years back of the envelop and very rough $0.5bn upwards. The hidden value creation might even be bigger like $6bn within the last 6 years so $1bn/year might be another number. That's huge!
- Another way to look at it is this: the value creation of wholly owned businesses is always inflated in the books. And this part is growing a lot within Fairfax (and before at Markel and Buffett) within the last years. So one could argue that "intrinsic equity" is bigger than "Face equity". And the hidden part of the intrinsic equity is creating intrinsic (often hidden) earnings every year. Is it understandable what I mean?


If we may agree that Prem is able to earn ROE of 13% on the investment portfolio, than what do we both mean with that number?

I can say, what I mean: For me that is just another way of saying "I expect Prem to outperform the value creation of the stock market by a few percentage points per year on average with his investment decisions on the equity investment portfolio."

And I base this expectation of "better than average value creation" on the true intrinsic value as a starting point. Not just 13.2% of inflated 'face equity'. But 13.2% also on the hidden assets. 

And I am pretty sure that in the case of FFH the equity investment portfolio is bigger than the $20bn FFH had end of year "in their books" So if we assume ROE of 13.2% to be the average for the next e. g. 10 years, at least what I think is: "Yes, but I don't expect that 13.2% to happen on the $20bn face value of the investment portfolio (which would be 2.64bn earnings/year and which is the basis of the official numbers in the tables), but I expect Prem to get the 13.2% returns on $26bn+ (so more like 3.43bn+ of earnings/year). If I put back that 3.43bn, that would be a return of 17.2% on the "face value" of the $20bn of the equity investment portfolio.

I am not sure, if that makes sense; I just find more and more hidden aspects, that makes me think, there might be more to it. 

 

BTW: What do you think Prem wants to say, when comparing last years investment return to the average of the 25 years? Is that just a random comparison, that came to his mind? Taking the average of a random number of random years of the past? Or is it dedicated, but this is just like Prem is saying "Oh wow, I didn't expect that returns of the first 25 years to happen again ever! I always thought it would be impossible to get back to that early investment portfolio returns ever. Not even in a single year. And now it happened - I can't believe it's true. But please don't expect that to come back ever! It's just a random onetime coincidence, something that's definitely never to come back again." - but where's the disclaimer than for the future - if it would be a onetime thing in Prems eyes, I am pretty sure he would have addressed it - but he hasn't. Or what else could it mean? I just don't find any other interpretation than "Okay, we are on a good track - let's see, if we can reach something similar again" The latter interpretation to me makes sense and at least I can't find any other reasonable interpretation of what the statement might otherwise mean.

[dartmonkey]

14 hours ago, SafetyinNumbers said:

 

14 hours ago, dartmonkey said:

I calculate the last line by backing out the fixed income return from the overall return

Do you include gains/losses in the bond portfolio in the fixed income return?

 

No, I kept it with non-FI investing gains. I thought about putting it with fixed income, since it is from trading fixed income securities, even if it is clearly not part of the 5.0% fixed income return.

 

Perhaps I will switch it, since it is clearly reported in the annual statements and it doesn’t really make sense to have bond sale gains in the numerator compared to non-fixed income investment assets in the denominator. What do you think?

[SafetyinNumbers]

58 minutes ago, dartmonkey said:

No, I kept it with non-FI investing gains. I thought about putting it with fixed income, since it is from trading fixed income securities, even if it is clearly not part of the 5.0% fixed income return.

 

Perhaps I will switch it, since it is clearly reported in the annual statements and it doesn’t really make sense to have bond sale gains in the numerator compared to non-fixed income investment assets in the denominator. What do you think?

It’s not necessarily from trading, just the change in interest rates. It understates the equity returns by having it included there. 

[TwoCitiesCapital]

FFH up 6% in the last 5-days vs S&P down ~1% over the same period. 

 

Seems like Fairfax NCIB doing its work. Maybe another 2022 like year coming where it's up significantly while indices are down. 

[MungerWunger]

6 minutes ago, TwoCitiesCapital said:

FFH up 6% in the last 5-days vs S&P down ~1% over the same period. 

 

Seems like Fairfax NCIB doing its work. Maybe another 2022 like year coming where it's up significantly while indices are down. 

 

[UK]

43 minutes ago, TwoCitiesCapital said:

Maybe another 2022 like year coming where it's up significantly while indices are down. 

And if not, more shares to buy back, win-win

[SafetyinNumbers]

1 hour ago, TwoCitiesCapital said:

FFH up 6% in the last 5-days vs S&P down ~1% over the same period. 

 

Seems like Fairfax NCIB doing its work. Maybe another 2022 like year coming where it's up significantly while indices are down. 

I think we must have picked up an institutional buyer off of Poseidon. NCIB can only buy on downticks. 

[mananainvesting]

44 minutes ago, SafetyinNumbers said:

I think we must have picked up an institutional buyer off of Poseidon. NCIB can only buy on downticks. 

Damm, I did not know this!! 

Question: Fairfax has been buying shares via NCIB over the years at increasing prices, how is that possible? (unable to get a clear answer from Chatgpt)

[dartmonkey]

53 minutes ago, mananainvesting said:

Question: Fairfax has been buying shares via NCIB over the years at increasing prices, how is that possible? (unable to get a clear answer from Chatgpt)

If there is a no-uptick rule, this only means the previous trade has to be at a lower price, it doesn't mean you can't repurchase shares when the general trend is upwards, as it has been for 3.5 years (I was going to say 5-6 years, but the share price in October 2022 was about $450, almost exactly the same price as in October 2014... 

[dartmonkey]

6 hours ago, SafetyinNumbers said:

It’s not necessarily from trading, just the change in interest rates. It understates the equity returns by having it included there. 

Yeah, I had thought it was more of a trading thing but I guess they add even the unrealized gains every year and it adds a lot of noise. The average in the last 6 years is close to 0 - $19m in fact, so I just excluded it, since adding it to interest and dividends returns from bonds would mean the bond returns are all over the map, whereas the bond yield is the one thiing that is pretty consistent from year to year. I think it is reasonable to expect that on average the market gains and losses will be about zero, which is maybe slightly on the conservative side, since they do trade in and out occasionally based on macro guesses, but for the moment I'll just leave those gains or losses on bonds out of the model. As expected, it makes the equity returns less variable:

 

[SafetyinNumbers]

31 minutes ago, dartmonkey said:

Yeah, I had thought it was more of a trading thing but I guess they add even the unrealized gains every year and it adds a lot of noise. The average in the last 6 years is close to 0 - $19m in fact, so I just excluded it, since adding it to interest and dividends returns from bonds would mean the bond returns are all over the map, whereas the bond yield is the one thiing that is pretty consistent from year to year. I think it is reasonable to expect that on average the market gains and losses will be about zero, which is maybe slightly on the conservative side, since they do trade in and out occasionally based on macro guesses, but for the moment I'll just leave those gains or losses on bonds out of the model. As expected, it makes the equity returns less variable:

 

That’s helpful. I think it shows the floor for equity returns might be higher because of the equity accounted for positions. On a related note, I thought it was interesting that consolidated positions aren’t in the investment returns they list. 

[dartmonkey]

20 hours ago, Hamburg Investor said:

I've got some questions to you:
- I go with you, that 18.8% is nothing I would expect on average to be reachable. More like 12% or maybe a bit more over time on average. Maybe the 13% is a good number to start with. So ROE 13% on the equity part.

 I'll answer the questions one by one, if I get to them all! But just to be clear, when I exclude the unrealized gains on the bond portfolio (which average out to about zero anyways), I get an average 13.6% on equity investments over the past 6 years, not too different from the initial estimate of 13.2%. Last year's 17% was maybe a bit of an outlier. There is one additional step in the RBC analysis of iinsurance companies, where they look at how much of an impact those gains have on ROE. This obviously depends on how much equity investments are in comparison with the company's book value (also called equity, confusingly...)

 

So for instance last year's 17% return on their $21,139.9 worth of non-fixed income investments was $3,584.2m, and that, in relation to their average book value last year, $24,621.2, gives them a 14.1% effect on ROE, i.e. that 14.1% gets added to the effect of underwriting on equity (7.4%; i.e. the 7% they make on underwriting, times the 1.07x leverage they get from having 1.07x as much net premiums written as average equity), and so forth. The total ROE I get for last year after tax would be 22.7%, which seems a little high. And assuming a drop of their combined ratio to 0.94 next year (along with stagnant net premiums written, with the softer insurance market) and an average 13.6% return on non-FI investments (not quite as good as last year's 17%, although we are off to a great start), I would get an 18.1% ROE this year, instead of 22.7%.

[dartmonkey]

20 minutes ago, SafetyinNumbers said:

On a related note, I thought it was interesting that consolidated positions aren’t in the investment returns they list. 

I think the consolidated non-insurance earnings have historically been small enough that they been lumped in with corporate overhead and interest and just disappear:

[Hamburg Investor]

20 hours ago, dartmonkey said:

So for instance last year's 17% return on their $21,139.9 worth of non-fixed income investments was $3,584.2m, and that, in relation to their average book value last year, $24,621.2, gives them a 14.1% effect on ROE, i.e

How did you arrive at $21bn? I may be a bit off the mark, but in the 2024 shareholder letter, the stock investments were stated as $17,5bn (CV) and $19bn (MV). Does the difference come from the TRS?

I’m already curious to hear your thoughts on the point regarding ROE and the intrinsic value of the investment portfolio. I see the key point in the difference between the CV and the actual intrinsic value of the equity portfolio. My base case is, that a fairly valued equity investment portfolio from Prem will outperform the stock market (10%) by a bit (2%+). At least in the very long term, i.e. before any withdrawals or reallocations into the stock portfolio.

[dartmonkey]

2 hours ago, Hamburg Investor said:

How did you arrive at $21bn? I may be a bit off the mark, but in the 2024 shareholder letter, the stock investments were stated as $17,5bn (CV) and $19bn (MV). Does the difference come from the TRS?

 

 

First of all, you are looking at the wrong year: last year was 2025, not 2024, and the equivalent numbers for carrying value and market value are up to $20.5b and $23.6b, iin the table on p.16

 

But I actually used another table, giving the carrying values of common stocks, on p.179 (of the 2025 AR):

 

Thinking about this further,

(a) I don't know why the two numbers, $21,139.9b and $20.5b, are not the same, and 

(b) Since I am interested in separating invesment results into fixed income and non-fixed income, I really should be adding the returns from preferred stocks and real estate, not just common stocks. I'm just not sure that the line item 'interest and dividends' only pertains to fixed income interest and dividends; does anyone know? For that matter, some of the common stocks issue dividends, too, so if they are included in 'interest and dividends' (for instance, as below), I don't know how I can isolate fixed income, except by using their yield on fixed income (5.0% last year) and estimating how much that represents, based on the size of the fixed income portfolio. If anyone has any thoughts about this, they would be welcome.

 

Here's the extract from p.17 whose wording makes me feel that when they refer to interest and dividends, they are just talking about fixed income dividends, not common stock dividends:

 

[Hamburg Investor]

33 minutes ago, dartmonkey said:

irst of all, you are looking at the wrong year: last year was 2025, not 2024, and the equivalent numbers for carrying value and market value are up to $20.5b and $23.6b, iin the table on p.16

But the 2025 stock returns were happening on the basis of year end stock investment portfolio of 2024, weren’t they? So investment portfolio ROE for 2025 takes the returns of 2025 and puts them in relationship to ye book of 2024. Or am I off here?

 

33 minutes ago, dartmonkey said:

(b) Since I am interested in separating invesment results into fixed income and non-fixed income, I really should be adding the returns from preferred stocks and real estate, not just common stocks. I'm just not sure that the line item 'interest and dividends' only pertains to fixed income interest and dividends; does anyone know? For that matter, some of the common stocks issue dividends, too, so if they are included in 'interest and dividends' (for instance, as below), I don't know how I can isolate fixed income, except by using their yield on fixed income (5.0% last year) and estimating how much that represents, based on the size of the fixed income portfolio. If anyone has any thoughts about this, they would be welcome.

 

I‘ve got the same problem. It would just make so much more sense to me to separate income from equity and from insurance/bonds. I am not an accountant, but I would intuitively like to put the whole stock divs and add those to the equity portfolio to get to a solid number regarding the real equity portfolio returns (I mean: Divs are an important part for understanding the returns and Prems ability in stock investment. And intuitively I‘d like to add the fixed income from bonds etc. to the insurance earnings (or losses).

That way we‘d have two clearly separated and consistent baskets.

33 minutes ago, dartmonkey said:

 

Edited March 19 by Hamburg Investor

[mengan]

From the Proxy statement:

 

From the annual report:

 

Am I correct to read that they cancelled 206,515 shares by 06.03.26 compared with 31.12.25?

[dartmonkey]

15 hours ago, Hamburg Investor said:

But the 2025 stock returns were happening on the basis of year end stock investment portfolio of 2024, weren’t they? So investment portfolio ROE for 2025 takes the returns of 2025 and puts them in relationship to ye book of 2024. Or am I off here?

I actually used the average of year-end 2024 and year-end 2025 as the denominator, which I call “year average fixed income portfolio”, but you are right, 2025 year-end value is not relevant. 

[Hamburg Investor]

9 minutes ago, dartmonkey said:

I actually used the average of year-end 2024 and year-end 2025 as the denominator, which I call “year average fixed income portfolio”, but you are right, 2025 year-end value is not relevant. 

Okay, so you need to replace $21,139m with $16,930m (that's the number from 2024). In that case, wouldn’t the equity portfolio have grown by 21.2% rather than 17%, with earnings at $3,584.2m?

I must admit that I also find it difficult to find the right size here. 

Maybe someone else here could help us? 

[Viking]

Fairfax's $50B fixed income portfolio is positioned very conservatively:

• Low average duration
• Mostly in government bonds

6 months ago there was a lot of hand wringing about falling rates and what this would do Fairfax's interest income. 

 

Bond yields across the curve are spiking. The 2 year Treasury yield is up about 50 basis points over the past 3 weeks. This will give Fairfax the opportunity to extend the average duration of their fixed income portfolio.

 

Interest and dividend income is Fairfax's largest income stream. Extending the average duration would lock this income stream in at a very high level ($2.5B) for a few more years.

 

Bottom line, Fairfax's fixed income portfolio is perfectly positioned for the current environment.

Edited March 20 by Viking

[Hoodlum]

15 minutes ago, Viking said:

Fairfax's $50B fixed income portfolio is positioned very conservatively:

• Low average duration
• Mostly in government bonds

6 months ago there was a lot of hand wringing about falling rates and what this would do Fairfax's interest income. 

 

Bond yields across the curve are spiking. The 2 year Treasury yield is up about 50 basis points over the past 3 weeks. This will give Fairfax the opportunity to extend the average duration of their fixed income portfolio.

 

Interest and dividend income is Fairfax's largest income stream. Extending the average duration would lock this income stream in at a very high level ($2.5B) for a few more years.

 

Bottom line, Fairfax's fixed income portfolio is perfectly positioned for the current environment.

thanks @Viking  do we know what the duration was at year end.  I don’t remember seeing that in the annual report. 

[Viking]

1 hour ago, Hoodlum said:

thanks @Viking  do we know what the duration was at year end.  I don’t remember seeing that in the annual report. 

 

Fairfax did not tell us on the Q4 conference call (they usually do). I took this to mean that it was lower. I think the last number we got from the company was 2.4 years.

 

If current trends continue in financial markets, there are going to be a lot of puts and takes when Fairfax reports Q1 results:

• Equities sell off
• Spike in bond yields 

I will update my equity tracker at quarter end.

 

Volatility is a very good thing for Fairfax. For example: the spike in bond yields will allow Fairfax to extend duration, which will lock in $2.5B in interest and dividend income for a few more years. But it will result in a sizeable loss in fixed income (offset by a large gain due to IFRS - not sure of magnitude). So there will be some short term noise. But long term gain.

Edited March 20 by Viking

[Hoodlum]

37 minutes ago, Viking said:

 

Fairfax did not tell us on the Q4 conference call (they usually do). I took this to mean that it was lower. I think the last number we got from the company was 2.4 years.

 

If current trends continue in financial markets, there are going to be a lot of puts and takes when Fairfax reports Q1 results:

• Equities sell off
• Spike in bond yields 

I will update my equity tracker at quarter end.

 

Volatility is a very good thing for Fairfax. For example: the spike in bond yields will allow Fairfax to extend duration, which will lock in $2.5B in interest and dividend income for a few more years. But it will result in a sizeable loss in fixed income (offset by a large gain due to IFRS - not sure of magnitude). So there will be some short term noise.

 

That was quite the spike today for all treasury terms.  We may see 5% again for the 30 year very soon.  I believe this rise in yield is mainly due to the war but I don't think the bond or equity markets have fully anticipated the downstream impacts of the huge increases in Memory prices over the past 6-12 months.  This impacts everything from consumers devices, vehicles, Security/networking and hosting services.   

 

I have started to see this with security/network device deals that are getting push out by months on previously agreed on deals as pricing is out of whack now.  The choice for the business is to pay more now to get the product, cancel order or delay for 12-18 months and hope that prices come down next year.  Anyone hosting services will start to see big monthly increases which then needs to be passed on.

 

This is going to cause inflation, along with companies pulling back on planned projects involving hardware/service upgrades.  The increase in the oil price only adds to this.

[dartmonkey]

6 hours ago, Hamburg Investor said:

Okay, so you need to replace $21,139m with $16,930m (that's the number from 2024). In that case, wouldn’t the equity portfolio have grown by 21.2% rather than 17%, with earnings at $3,584.2m?

I do use earnings of $3584.3m (was 3584.2 a typo or maybe I have a typo somewhere? I have $3,967.5 total gains, and subtract off 2025's $383.2m in bonds gains) in the numerator. In the denominator, it seems to me more logical to use the average non-fixed income portfolio for 2025, not the beginning of the year and not the end of the year, so I just use the straight average of end of 2024 and end of 2025.

 

For common stocks, that would be $16,930m and $21,139m, but since I really just want to separate out fixed income (cash, cash equivalents, short-term investments and bonds) from the rest, I also lumped in preferred stocks and real estate along with common stocks, and took the average of these values at the end of 2024 and the end of 2025, so I ended up getting $22,306.8m as my denominator, for a return of $3584.3m/$22,306.8m = 16.1% for non-FI for 2025. The average for the last six years was 12.0%.

 

 

It's a work in progress, but I'm getting there...