Fairfax 2023

[Viking]

14 hours ago, nwoodman said:

@Viking Great work as always.  Agree capital allocation/deal flow is key moving forwards.  Higher for longer definitely helps though, if nothing else it keeps the proverbial foot on the competition’s throat.  Analysts forecasts are still very much in the “they screw it up” and interest rates fall fast.  FWIW (not much) this is the view of the few: FY23 $US135, FY24 $US118 and FY25 $US79 they then fall of the proverbial cliff (our Morningstar friend no doubt).  A very different view indeed.

 

In terms of capital allocation, I see a minimum 10% just via Omer’s buybacks, then around 12-15% buying back their own shares.  So they should have no problems reallocating incremental capital at sensible rates at least for the next few years.  Even then,   I don’t believe that the 11-12% compounding machine, that has been in operation for 30+ years, is broken. You have to keep telling yourself that if these “analysts” were any good they would be running their own book.

 

@nwoodman my guess is analysts struggle with the volatility in Fairfax’s earnings. The industry views volatility as ‘risk’ and as a big negative (hence, big hair cut to earnings). Which just seems a little bizarre to me. 
 

i think analysts also struggle with the current size of the spike in earnings. They don’t trust its staying power - lots of posters on the board also don’t trust the sustainability of earnings. 
 

So estimates for 2024 are lower than 2023. And estimates for 2025 are lower than 2024 (much lower than 2023). Makes no sense to me. 

Edited October 8, 2023 by Viking

[Maverick47]

To be fair, the estimates through 2024 are an average of 6 analysts, while for the years thereafter, only one analyst provided EPS estimates.

[SafetyinNumbers]

I think part of is it they miss the IFRS 17 impact on earnings because they model underwriting income on the stated combined ratio while the IFRS adjusted combined ratio has been lower. Viking has something in line 4 of his model above. When I look at RBC’s model, they have the combined ratio down but also have underwriting income down year over year. When I was in equity research we would say that’s not internally consistent.

 

For 2024, the growth in investment income is underestimated based on current rates, associates income is held flat from 2022 despite H123 at ~60% of 2022 already and gains expected on the equity portfolio are very small. He’s at $130/sh for 2024 and he might be right but the odds seem low.
 

 

[nwoodman]

15 minutes ago, SafetyinNumbers said:

I think part of is it they miss the IFRS 17 impact on earnings because they model underwriting income on the stated combined ratio while the IFRS adjusted combined ratio has been lower. Viking has something in line 4 of his model above. When I look at RBC’s model, they have the combined ratio down but also have underwriting income down year over year. When I was in equity research we would say that’s not internally consistent.

 

For 2024, the growth in investment income is underestimated based on current rates, associates income is held flat from 2022 despite H123 at ~60% of 2022 already and gains expected on the equity portfolio are very small. He’s at $130/sh for 2024 and he might be right but the odds seem low.
 

 

Any chance you could post the RBC model?

[SafetyinNumbers]

3 minutes ago, nwoodman said:

Any chance you could post the RBC model?

 

[nwoodman]

59 minutes ago, SafetyinNumbers said:

 

 

Thanks @SafetyinNumbers.  What is RBC's price target?  Taking these numbers or @Viking great work, this still seems very cheap.  The current price surely is predicated on capital destruction at some point, either through misallocation of capital, seizure of foreign assets or massive policy misplacing.

 

Running a 20 year test on P/B.  The mean for Markel, Berkshire, Fairfax is:

 

MKL 1.5x

BRK 1.4x

FFH. 1.0x

 

Nothing new in the observation that long ROE is what will drive share price.  Over the last 20 years Mean ROE has been

 

MKL 7.9%

BRK 9.3%

FFH 8.1%

 

I totally agree with the observation that for a decade of this period, the float was not adding meaningfully to ROE, that's changed.  A couple of years of +1 STD deviation (18%) in FFH's ROE seems very likely.  Whether this results in +1 standard deviation of 1.3x's book I am not sure, but it seems like an asymmetric bet.

 

 

Edited October 9, 2023 by nwoodman

[SafetyinNumbers]

7 minutes ago, nwoodman said:

Thanks @SafetyinNumbers.  What is RBC's price target?  Taking these numbers or @Viking great work, this still seems very cheap.  The current price surely is predicated on capital destruction at some point, either through misallocation of capital, seizure of foreign assets or massive policy misplacing.

 

Running a 20 year test on P/B.  The mean for Markel, Berkshire, Fairfax is:

 

MKL 1.5x

BRK 1.4x

FFH. 1.0x

 

Nothing new in the observation that long ROE is what will drive share price.  Over the last 20 years Mean ROE has been

 

MKL 7.9%

BRK 9.3%

FFH 8.1%

 

I totally agree with the observation that for a decade of this period, the float was not adding meaningfully to ROE, that's changed.  A couple of years of +1 STD deviation in FFH's ROE seems very likely.

 

 

They are at US$980. 
 

The increase in interest rates means a structural increase in ROE vs the last 20 years. Should be interesting what the narratives will be when the multiple expands and how quickly holders will jump ship. There will surely be lots of drawdowns that investors will want to avoid. 

[StubbleJumper]

28 minutes ago, nwoodman said:

 A couple of years of +1 STD deviation (18%) in FFH's ROE seems very likely.  Whether this results in +1 standard deviation of 1.3x's book I am not sure, but it seems like an asymmetric bet.

 

It's absolutely an asymmetric bet. 

 

Anyone can take @Viking's forecasts for the current year and the next two and develop optimistic, pessimistic and mid-point scenarios for BV on Dec 31, 2025.  Even if you give a haircut those forecasts out of an abundance of caution to create a pessimistic scenario, it's hard to envision a scenario where BV isn't US$1,100 by year-end 2025.  With last week's market price of ~US$850, a buyer last week would have a perfectly nice return over the next couple of years without any P/BV expansion at all, and that's based on a somewhat pessimistic scenario that applies a modest haircut to @Viking's forecasts.

 

If get a bit braver and ignore the desire for an abundance of caution and accept @Viking's forecasts as they've been presented, it's even better.  And then if you get really, really ballsy and dream about an outrageous P/BV of 1.1 on Dec 31, 2025 you get an outstanding return.  You don't need 1.3x or 1.4x BV to get a great result over the next couple of years from an investment today.

 

So yeah.  Asymmetric is exactly the right description.  If things go slightly poorly, you get a perfectly acceptable return, and if the moons and stars are even slightly aligned it could be considerably better 

 

SJ

[giulio]

FFH floating rate preferred shares are trading at >10% yield, possible redemption in dec 24/dec 25.

Canadian 3 month t-bill rate + spread (2.5-3%), resetting quarterly.

Bradstreet bought these in dec 2022.

The probability that they get called is low IMO, still looks like a good deal.

What am I missing? 

[MMM20]

2 hours ago, giulio said:

FFH floating rate preferred shares are trading at >10% yield, possible redemption in dec 24/dec 25.

Canadian 3 month t-bill rate + spread (2.5-3%), resetting quarterly.

Bradstreet bought these in dec 2022.

The probability that they get called is low IMO, still looks like a good deal.

What am I missing? 

 

Are those prefs a better risk/reward than the common at a ~18-20% earnings yield? I agree they look good (especially vs. FFH's fixed rate prefs @ ~6%) but I think Fairfax has a long runway for ROIC > WACC (most of which, again I'm a broken record, is float-based leverage at ~0% cost) so I still want 100% common.

 

Edited October 9, 2023 by MMM20

[StubbleJumper]

2 hours ago, giulio said:

FFH floating rate preferred shares are trading at >10% yield, possible redemption in dec 24/dec 25.

Canadian 3 month t-bill rate + spread (2.5-3%), resetting quarterly.

Bradstreet bought these in dec 2022.

The probability that they get called is low IMO, still looks like a good deal.

What am I missing? 

 

The prefs are great for certain people.  If you are a Canadian tax filer with a modest taxable income, in certain circumstances, you might end up paying no income tax at all on the ~10% dividend, which makes for a pretty attractive net return.  And then if interest rates should happen to decrease, you could also be in line for a modest capital gain to top things off.   So, they are definitely worth looking at, especially for a particular subset of Canadian taxpayers.

 

The prefs do, however, have a few issues. The market is dominated by unsophisticated retail investors so when the market gets a little choppy, the prefs can experience irrational price swings.  If you happen to need to repatriate your capital during one of those irrational periods, you could face a capital loss.  Secondly, the dividends reset every five years, so the double-digit return might be time limited if interest rates decline.  Thirdly, a foreigner will be subject to a considerable withholding tax on the dividends, unless Canada has a tax treaty with the foreign country that might reduce the withholding tax.  Fourthly, the prospectus for the prefs allows FFH to redeem them when it is attractive for the company, and those situations are generally not attractive for the investor, but there is no retraction provision allowing the investor to retract them with conditions are favourable for the investor (and unfavourable for FFH).  Finally, as with all prefs, you are subject to risk of financial failure by FFH and you might not be adequately compensated for that risk because your upside is capped with a pref (Let's be honest with ourselves.  Prem seems comfortable with significant financial leverage and has come close to driving FFH into a wall on a couple of occasions).

 

A Canadian investor looking to exploit our income tax provisions for dividends paid by Canadian Controlled Private Corporations might look at FFH prefs and be attracted at this stage.  But he should equally be looking at the common shares of our chartered banks, as outfits like the Canadian Imperial Bank of Commerce and the Bank of Nova Scotia currently sport 7% dividend yields and those divvies are bumped up a shade twice per year.  At the end of five years (the FFH reset period) it is likely that the yield on purchase price for one of those banks would be pretty similar to the 10% yield on the FFH prefs.  And those banks are the most liquid of Canadian equities and face virtually no risk of financial failure by the underlying company.

 

At this point, the prospects for FFH common shares over the next few years are so attractive that someone who doesn't have a specific income tax motivation would probably just buy the common instead of the FFH prefs.  A foreigner with no income tax advantage would almost certainly buy the common....

 

 

SJ

[giulio]

@MMM20, @StubbleJumper thanks for the feedback!  

The relative attractiveness of the subordinate shares over the preferreds is what makes me think the probability of redemption is low.

I looked at them but decided to pass since I would be double-taxed both in Canada and Italy.

 

On to the next idea!

 

 

 

 

 

 

[UK]

4 hours ago, MMM20 said:

 

Are those prefs a better risk/reward than the common at a ~18-20% earnings yield?

 

You know, it is very good question to ask and not only vs preferreds. But if one to agree with you yield asumption (which I more or less do), it is really hard for something else to compete with it (or even to clear this hurdle). Even such seemingly cheap stocks, like M or C or something from Oil and Gas, or you name it, they are more or less as cheap, yet I would argue, that FFH is of a much higher quality and much better for a long term holding. 

[Thrifty3000]

On 10/8/2023 at 12:36 AM, Viking said:

4.) Effects of discounting and risk adjustment (IFRS 17).

• Interest rate changes drive this bucket. My estimates here could be a little messed up. Given I am forecasting interest rates to remain about where they are today, I am leaving this number the same over the forecast period (at my estimate for June 30, 2023).

@Viking why are you adding $480 annually for this adjustment? If you're assuming interest rates remain flat wouldn't that make the annual adjustment for this $0? And, wouldn't that reduce the EPS by around $20 per share in 2024 and 2025? 

[SafetyinNumbers]

2 hours ago, Thrifty3000 said:

@Viking why are you adding $480 annually for this adjustment? If you're assuming interest rates remain flat wouldn't that make the annual adjustment for this $0? And, wouldn't that reduce the EPS by around $20 per share in 2024 and 2025? 

I know you asked Viking but I think about this a lot so I hope you don’t mind my thoughts.
 

My understanding with IFRS 17 is as long as interest rates aren’t zero there will be some sort of adjustment. The reported combined ratio does not include any impact for discounting reserves. But every quarter, the existing reserve balance accretes and any reserves for new policies have to be discounted. If rates are flat or going up, that should be a sizeable benefit every quarter. If rates are going down, the reserve balance will be revalued higher but the discounting of the new policies will still be positive. 
 

I think most analysts are ignoring this and that’s part of why their earnings estimates are too low. Intact breaks out the discounted combined ratio (see below) and for them in the first half it was a 440bps difference. I’m not sure what the right number is for Fairfax but it’s not zero. That being said at some point in the future if rates fall fast enough, the discounted combined ratio might be higher than the reported combined ratio.

 

[sleepydragon]

why is Fairfax down today? is there a downgrade or something? or is it Tbill yield is down?

[Santayana]

I wouldn't bother trying to understand short term moves in Fairfax stock.  But maybe just some profit taking after hitting ATHs on Friday?

Edited October 10, 2023 by Santayana

[Thrifty3000]

2 hours ago, SafetyinNumbers said:

I know you asked Viking but I think about this a lot so I hope you don’t mind my thoughts.
 

My understanding with IFRS 17 is as long as interest rates aren’t zero there will be some sort of adjustment. The reported combined ratio does not include any impact for discounting reserves. But every quarter, the existing reserve balance accretes and any reserves for new policies have to be discounted. If rates are flat or going up, that should be a sizeable benefit every quarter. If rates are going down, the reserve balance will be revalued higher but the discounting of the new policies will still be positive. 
 

I think most analysts are ignoring this and that’s part of why their earnings estimates are too low. Intact breaks out the discounted combined ratio (see below) and for them in the first half it was a 440bps difference. I’m not sure what the right number is for Fairfax but it’s not zero. That being said at some point in the future if rates fall fast enough, the discounted combined ratio might be higher than the reported combined ratio.

 

 

Ok, but if you go back to @Viking's latest model/explanation you can see the same $480 mil added to estimated income every year (circled in red)...

 

 

My understanding is adding $480 mil each year would require you to assume interest rates would increase each year. However, in @Viking's commentary he states he is assuming interest rates will remain flat...

 

 

Therefore, it sounds to me like he should have added the $480 mil to 2023, and then assumed adjustments of $0 for that line item in 2024 and 2025, which would result in reduced per share earnings estimates of roughly $15 to $20 per year in those two years.

Edited October 10, 2023 by Thrifty3000

[SafetyinNumbers]

1 hour ago, Thrifty3000 said:

 

Ok, but if you go back to @Viking's latest model/explanation you can see the same $480 mil added to estimated income every year (circled in red)...

 

 

My understanding is adding $480 mil each year would require you to assume interest rates would increase each year. However, in @Viking's commentary he states he is assuming interest rates will remain flat...

 

 

Therefore, it sounds to me like he should have added the $480 mil to 2023, and then assumed adjustments of $0 for that line item in 2024 and 2025, which would result in reduced per share earnings estimates of roughly $15 to $20 per year in those two years.

I want to make it clear I’m not an insurance expert, my MAcc degree is 23 years old and I let my CA/CPA expire a few years ago to save on the fees!

My premise is that as long as interest rates are positive and rates are unchanged, the discounted combined ratio will be lower than the undiscounted combined ratio assuming a growing business. I assume when a policy is sold, premiums are collected and reserves are set aside. If those reserves are discounted, the underwriting profit is by definition higher all else being equal and that should happen every quarter. The quarterly offset, however, is the reserve balance must also accrete at the same discount rate. 
 

Before IFRS17, in order to model underwriting income, an analyst will most likely estimate a combined ratio based on the trend in the reported undiscounted combined ratio. After, IFRS17 that’s still all Fairfax is giving us explicitly so that’s still how underwriting income is being modelled. But there is a plug needed. I don’t know if $480m is a fair estimate.
 

If the discounted combined ratio is 300bps lower than the reported combined ratio and Fairfax writes $25b in policies, does that mean $750m in additional profits? In theory that includes any accretion from the reserve balance. I’m not sure at all but it makes sense to me. 

 

 

 

Edited October 11, 2023 by SafetyinNumbers

[Viking]

9 hours ago, Thrifty3000 said:

@Viking why are you adding $480 annually for this adjustment? If you're assuming interest rates remain flat wouldn't that make the annual adjustment for this $0? And, wouldn't that reduce the EPS by around $20 per share in 2024 and 2025? 

@Thrifty3000 I don’t think i can give a better explanation than @SafetyinNumbers has already provided. I would appreciate others providing their thoughts  - that CofBF collective wisdom thing. 
 

‘My estimate here could be a little messed up.’ Bottom line, i am still learning about this bucket. It will take me a few more quarters to better understand the build and see how this number evolves at Fairfax. As i learn more i will update my forecasts.

Edited October 11, 2023 by Viking

[UK]

2 hours ago, Viking said:

@Thrifty3000 I don’t think i can give a better explanation than @SafetyinNumbers has already provided. I would appreciate others providing their thoughts  - that CofBF collective wisdom thing. 
 

‘My estimate here could be a little messed up.’ Bottom line, i am still learning about this bucket. It will take me a few more quarters to better understand the build and see how this number evolves at Fairfax. As i learn more i will update my forecasts.

 

I think we already have discussed this before and I definitely can not claim that SafetyNumbers is not right on this, since I do not understand this myself completely (and who does?), but my initial understanding on this was in line with Thrifty3000's, meaning that large recurring gains from this item would be produced only if rates increase further substantially. But it seems it also depends on the change of net reserves, as SafetyNumbers has stated, so in the end it seems it depends on two variables: discount rate and net reserve change, and the final result for any period will be impacted by both. So I think this probably means, that such large initial benefit from applying this for the first time will not be repeated in the future, unless rates moves up substantially, despite of reserves growing at a steady pace. In such case (no rate change, reserves growing steady), my guess, the impact would still be positive, but way smaller?

 

From 2q CC: 

 

Under IFRS 17, our net earnings are affected by the discounting of our insurance liabilities and the application of a risk adjustment. In the second quarter of 2023, our net earnings benefited $221 million pre-tax from the effects of discounting losses occurring in the current quarter, changes in the risk margin, the unwinding of the discount from previous years and changes in the discount rate on prior year liabilities. As interest rates move up and down, we will see positive or negative effects on earnings from discounting.

 

From 1q CC:

Tom MacKinnon

Great. Yes, Jen, I was just wondering, the things that really impact IFRS 17, the change in the risk adjustment, the unwind of the discount, the build of the discount and the change in the discount rate. So, if we kind of had a flat interest rate environment and pretty well steady state with respect to your growth. Would all of this noise be pretty minimal, like what kind of conditions would make this noise show up more to the positive or actually show up more to the negative?

Jennifer Allen

Yes. Sure. It's a good question, Tom. So, the way I think if you're in a steady state, if your underlying net reserves from a risk profile duration does not change, then as you unwind your discounting that you don't have a change in your discount rate, it should really be offset and really don't see a huge impact. The other side of it is, your risk adjustment would be steady state, you would be releasing your risk adjustment on your old book, but you would also be setting up the same risk adjustment on your new book. So it's only when your book grows, so if your net reserve starts to grow, you'll start to get that net benefit through again, if it shrinks, it would be a negative impact to your total portfolio.

Tom MacKinnon

And then on the change in the discount rate, is that just generally, if we have a flat interest rate environment, then we wouldn't get that noise as well, I assume.

Jennifer Allen

Correct.

 

 

[giulio]

The accounting changed but the economic substance did not. I think it is not worth it to focus on the impact of discounting/interest rates. it will move up and down, like mark-to-market investment gains (which should be excluded from earnings and not projected forward). 

I try to measure look-through earnings instead.

I try to understand if they are writing profitable business and what CR will be over a cycle.

I think Mr. Watsa said on CC that IFRS 17 will not have any impact on the way FFH conducts business, so let the equity research analysts deal with this mess in their models!

focus on the business, don't let the accounting obfuscate economic reality.

 

[sleepydragon]

reading Fairfax's annual letters last night. I am new to this stock.. but they are writing insurance everywhere in the world -- like Indonesia etc.. Do they really know what risks they are taking on? It makes me a bit hard to sleep at night. 

[UK]

41 minutes ago, giulio said:

The accounting changed but the economic substance did not. I think it is not worth it to focus on the impact of discounting/interest rates. it will move up and down, like mark-to-market investment gains (which should be excluded from earnings and not projected forward). 

I try to measure look-through earnings instead.

I try to understand if they are writing profitable business and what CR will be over a cycle.

I think Mr. Watsa said on CC that IFRS 17 will not have any impact on the way FFH conducts business, so let the equity research analysts deal with this mess in their models!

focus on the business, don't let the accounting obfuscate economic reality.

 

 

Well, I agree, but if you still want to model any future EPS/BV, as Viking does, you have to decide what to do with this item:). Perhaps to stay conservative one can omit it altogether. Look through earnings is an alternative/different way to look at investment portfolio,  I agree it may be better in some respects, but maybe not for EPS estimate.

 

Edited October 11, 2023 by UK

[StubbleJumper]

7 minutes ago, sleepydragon said:

reading Fairfax's annual letters last night. I am new to this stock.. but they are writing insurance everywhere in the world -- like Indonesia etc.. Do they really know what risks they are taking on? It makes me a bit hard to sleep at night. 

 

 

When they write policies in Indonesia or Brazil, they are not doing so from headquarters in Toronto!  It is almost always a local subsidiary that does the underwriting in those countries.  In general, those local subsidiaries have been in operation for many years (decades!) and know the situation on the ground very well.  FFH has bought several of those insurance companies around the world.

 

There is an argument that owning several foreign subs might make you actually sleep better.  Something bad might happen in Indonesia, but if that occurs, it's unlikely to be accompanied simultaneously by bad news in South Africa, Brazil or Singapore.  The international subs are pretty geographically diverse.

 

 

SJ