dartmonkey

Yes, this is certainly a distinct possibility. As the S&P says in its methodology paper, "Within the S&P/TSX Composite, the S&P/TSX 60 covers large cap securities, with a view to matching the sector balance of the S&P/TSX Composite" (my emphasis.) As far as I can tell, it doesn't explicitly say that it is looking to include all the largest cap securities, just that they should be covered and match the Composite's sector balance. So one could argue that both CLS and particularly RBA would be better choices. At the moment, for the 3 sectors, as of August 29, we had Financials - In the Composite: 32.1%; in the 60: 36.8%, so the 60 is 4.7% overweight already, and more so with the loss of Teck, and the inclusion of FFH would make this about 1.5% worse; Information technology - Comp: 10.5%; 60: 11.4%, so 0.9% overweight already, and more with the loss of Teck, and the inclusion of CLS would make this about 1% worse; Materials* - Comp: 14.8%; 60: 12.2%, so 2.6% underweight. The loss of Teck, with its weight of about 0.7%, would mean that the 60 would be about 3.3% underweight, and the inclusion of RBA would make this about 0.8% better. So on this basis of this criterion, RBA would be the most logical pick, not FFH; they would be replacing a 'materials' company with another one, albeit smaller, but any other choice from the 3 biggest would make the weightings worse. (Is RB Global really a 'materials' business? It is basically an auctioneer, and this is how they describe themselves: "RB Global is your trusted global partner for insights, services, and transaction solutions for commercial assets and vehicles." But I don't think the S&P committee is going to get into the weeds about how much a 'materials' company really gets its hands dirty with actual materials.) If it really is going to take Investment Canada a year or more to decide whether they will approve of the Anglo American - Teck deal, there's a good chance some other opening in the 60 will crop up in the meantime, and for people hoping that Fairfax gets in, they should hope it's a financial company that is leaving, not a materials company.

From Yahoo FInance: "Anglo Teck would maintain its listings on the London and Johannesburg stock exchanges and also apply for listings on the Toronto and New York stock exchanges. The plan is to keep the company incorporated in London, which would mean the S&P/TSX composite index would lose Teck from its listings, since companies need to be based in the country to be included." Teck is the #49 Canadian company, by market cap. It is in the materials category of the S&P TSX Composite and the S&P, so if the deal goes through, which I expect it will, it would be dropped from both. It's market cap before today's 10% jump was about $24b, representing about 1% of the TSX 60, 0.8% of the ~210 companies on the TSX Composite and 0.6% of all ~1500 stocks traded on the TSX. Fairfax, at C$54b market cap, is currently the 25th biggest Canadian company, and by far the biggest not already included in the TSX 60. So it would seem likely that this might be its moment. It would skew the weight of the Financials in the TSX 60 a little farther from the overall Canadian market (currently 37% of the TSX 60, and 'only' 21% of the Composite), but what is the alternative? I suspect that is the reason FFH shares are up 2.2% today. But the S&P will probably wait for the merger to be approved before they toss Teck. The Globe thinks this might take Investment Canada a year and a half to decide (WTF?), so we had better be patient.

Good analogy. It depends on how you define ’sound’. But the trees falling in Fairfax’s forest are adding to the accumulated biomass, whether someone heard them or not. So when lightning eventually strikes, as it always does, sooner or later…

Yes. Check my logic, but the performance fee is good for Fairfax, not bad. For instance, a $2b gain in BIAL’s value means FIH has 74% of that, or $1.48b. Fairfax essentially gets all of its 43% share of this gain (80% accrual in value of FIH, plus 20% directly as a performance fee), PLUS 20% of the gain in value of the 57% of FIH shares held by other, non-FFH shareholders. In this example, that would be a pre-tax gain of 43%*$1.48b + 56%*20%*$1.48b boosting FFH’s gain from $636m to $802m.

Very nice analysis, thanks. And BIAL is indeed a nice example, along with Eurobank I think, of an asset whose true value is significantly higher than even 'fair value' (FV), which is itself higher than carrying value (CV). One quibble here, even though it wasn't included in your overall analysis, is that BIAL is 74% owned by Fairfax India, not Fairfax. Fairfax owns about 43% of Fairfax India, so Fairfax owns about 32% of BIAL, not 74%.

This is an excellent question, and a good follow-up to the discussion about ROE from last week. In the table that SafetyinNumbers posted, RBC Capital Markets said the return was 7.0%, with 2.8x leverage, implying a 19.9% contribution to ROE, in addition to underwriting gains (7.8%) but before overhead, financing drag, tax and preferred dividends, bringing the total of what they call operating ROE back down to 18.8% for 2024. This of course does not include the occasional realized gain or loss from opportunistic sales, which historically has provided additional return. But what is in that 7.0%? Given the end of 2024 breakdown of $17.5b in equity investments and $46.8b in the bond portfolio, plus a bit of cash and 'other', for a total investment portfolio of $67.4b, we have an equation where we can solve for the return of the equity investments, since we know the overall number (7.0%) and we also have the number Fairfax has reported (in the 2024 AR) for the bond portfolio (5.2%). That means we can calculated that the equity portfolio must have returned 13%, but as Viking asks, what type of return is that? Is it the reported earnings, using the unrealized gains in the mark to market portfolio, plus the dividends from associated investments, ignoring retained earnings until the investment is sold? I think the answer is yes, which would mean that any increase in fair value (FV) over carrying value (CV) is not in that earnings number. In addition, for some of their investments, one could legitimately ask whether 'fair value' is really fair. Eurobank is the best example - it is now trading at 7x earnings, but one might argue that it is really worth more than that, given its positioning, the strength of the Greek economy, the boost to earnings that will come as they integrate the Cyprus bank into their operations, etc. So hopefully there is an additional layer of true value (TV?) accumulating on top of FV, itself well above CV.

Seconded. Safety has posted this table before but I hadn’t appreciated how elegant it was. It clears up nicely the question about how much it is reasonable to estimate that underwriting contributes to ROE, and how much of a drag there is likely to be from overhead, financing, tax, and preferred shares (7.9% total, by my calculation.) Maybe there’s a companion table that breaks down how they come up with a pre-tax investment yield of 7.0%? At year end 2024 they had a bond portfolio of $46.8b earning 5.2% pre-tax, and a $17.5b stock portfolio, along with some cash, for a total investment portfolio of $67.4b. Assuming the cash is earning nothing, that would mean that to get to 7.0% pre-tax from investments (close to the long-term average of 7.7%), we would need $4.72b, with $46.8*0.052 = $2.43b of that coming from the bond portfolio, and the rest, $2.27b, coming from the equity portfolio. That would mean the equity portfolio would have to return $2.27/$17.5=13.0%. If they hit that mark, they would get an overall post-tax ROE of 18.8%, like in Safety’s chart. To hit their 15% ROE target, at a tax rate of 24.4%, they would need a pre-tax ROE of 15%/(1-.244)=19.8%, instead of the current 25.2% pre-tax, 5.4% lower. That would mean that if underwriting, interest and expenses, and bond returns stayed the same, equity investments would only need to return 14.4%, and given the 2.8x leverage, that means we would only need 5.1% from equity investments. This seems almost impossible to miss, given the investments like Eurobank 6x earnings, so a yield of about 16-17% at year end, obviously a bit lower yield now) or Poseidon (carried at $1.9b, which was 7x 2024 earnings according to the annual report.) And this is without considering the possibility that Fairfax gets even better returns by selling opportunistically, like they have done recently with Stelco or the bond portfolio.

My chart was a summary of the change in the share price over the past 5 years of Fairfax’s large publicly traded equity holdings. I didn’t include sales like Stelco and Resolute Forest Products, both of which were exceptionally well timed and delivered significant value to shareholders. What are the equity decisions that Fairfax has made over the past 5 years that have not worked out? I mean big ones. I can’t think of any. I tend to agree with you that their decisions in the past 5 years have been very good, and I am not contesting that point. I am only pointing out that the test of how well recent decisions have worked out would be to make a list of those decisions, and the returns since then, not just looking at the biggest 20 positions today and how well they worked out. For instance, to take an extreme case, say they had had 20 $1b investments in 2019 that all went to $100m, and 20 $100m investments that all went to $1b. That would actually be a terrible result, with $22b in investments that went to $22b, i.e. zero annualized return. But your chart would show the 20 $1b positions and how brilliant they were, and the 20 bad decisions, which had only a small current value, would not appear on the chart. I am not saying that that is what has happened, just that there is a survivor effect that this sort of analysis is vulnerable to. What I don’t do is look at decisions they made in 2012 or 2013 or 2014 and extrapolate or try and draw any strong parallels to today. That was a lifetime ago. I am not sure what informational value it has anymore… other than they made a big mistake. And it appears they have learned from each of them. My view is Fairfax is a completely different company in 2025 than it was in 2013. I don't really agree - 10-12 years ago for me is not a lifetime ago. This was largely the same people who made a big mistake, and they could make the same mistake again. I think there are good reasons to believe they have learned the lesson, just as I hope I have learned my lesson not to short companies I think are hugely overvalued. I think their 40-year history of 17-18% annual increase in book value, with extreme lumpiness, is still relevant, and the bad period from around 2010 to 2020 is part of that. If you cut out that period, then you also lose some of the really good investments, like Stelco and Eurobank and Metlen. The beautiful thing about investing in value situations is that the bad decisions end up becoming a smaller part of the whole, so going forward, snafus like Blackberry are no longer a problem. But this was a very big investment, even 6 years ago, since holding onto it is also a decision. And so my main point is that a full accounting of their performance should include all the big capital allocations, not just the ones that are still worth a lot today.

Yes, I certainly agree, but if we're looking at recent results, say in the last 10 years, we should make sure we include the good and the bad, so if we just look at the investments that are worth the most right now, we miss on the bad investments that were big when they were made and have only become small now because of their bad performance (or because they were liquidated at a loss.) If you look at per share book value growth in the 10.5 y since the end of 2014, it is really a tale of two periods, a terrible first 5 years and a great subsequent 5-5.5 years. So ignoring the dividend ($10/y dividend, $15 in the last 2 years), book value per share went up 4.2% from the end of 2014 to the end of 2019, and then 16.9% from the end of 2019 to the end of 2024 (or 17.1% to the end of Q2 2025). Adjusted for the value of the dividends would add about another 2% to that, but results from 2015 to 2020 were still pretty bad, for a variety of reasons (BB, APR, Sandridge, hedges, soft insurance market, low interest rates) and great since then. It's encouraging to see those recent results return to the long term 18% average, and hopefully they will continue, but we should acknowledge that the company went through a very bad period not so long ago.

Yes, Go Digit General Insurance Limited (aka Digit Insurance, which owns Go Digit Life Insurance, a much smaller company) is worth close to $4b, so Fairfax's 49% stake (making it an equity-accounted associate) is worth close to $2b, if it were marked to market, putting it at #3 or #4 in terms of Fairfax's equity holdings.

Those results are impressive, but to play the devil's advocate, I think it's worth pointing out that the logic of the table, ranking investments according to their present value, tends to favour the investments that have done well and not the ones that have done poorly. For intance, if they had made a $1b investment in company DogCo (aka Blackberry), and it had gone to $150m in the last 5 years, it would not appear in the chart, whereas the company StarCo that went from $100m to $1b would appear towards the top, and the conclusion would be that all the big investments have done well in the last 5 years. As a reminder of the Blackberry story, they accumulated a 10% stake by 2013 for about $1b (does any have the exact price ?) and then offered $4.7b for the rest, an attempt which, thank God, was not successful. But they did invest another $1b in convertible debentures, which give them the right to purchase another big chunk of the company at $10 a share. Some shares were sold in 2013, and the debentures were converted in 2020, so the current arrangement is that they could acquire at $6 a share. At the current Blackberry share price of $3.60, those obviously don't have a lot of value. I do not have the patience to go over all the money Fairfax has invested in this sad story, but $1b going to $150-200m seems like a reasonable approximation, and would put the current value of the Blackberry stake outside the top 20, but clearly a very bad return for shareholders of a big investment.

Yes, I had forgotten Grivalia, which was actually a big part of the investment. The first shares they bought in 2014, E400m at a share price of E31, for 13.6% of the company. That investment almost went to zero - to E1.00, actually, when the company was forced to refinance by the Greek government in 2015. At that time, Fairfax invested another E350m at E1.00 (the refinancing), taking them to 17%; they also bought an 80% interest in Eurolife from Eurobank, for E315m, so this can not really be characterized as part of the Eurobank investment, unless you figure they got a sub-market price for Eurolife, which I doubt. By 2016, Eurobank's share priced had dropped as low as E0.30, so this was a serious money-loser at that point. Fast forward to 2018, when Fairfax sold its 51% stake in Grivalia to Eurobank for shares in Eurobank, taking its stake in Eurobank from 17% to 32%. At that point, Eurobank was trading at E0.62 per share, so this investment, almost half the total, was at an even better price than the E1.00 refinancing in 2015. That put the weighted average share price at E0.94. Given the fact that Fairfax has had serious money invested for about 10 years, and the share price is now E3.48 (another high today, although we will probably never see another all-time high), and there have only been 2 10c dividends, this can legitimately be said to have worked out fabulously well since 2020 (which is probably what Viking meant), compared to how terrible it was looking for the first 5 years, but for it to be a great investment overall, we probably need to see a little more of this greatness first.

As your note says, but not your table, it was 1.266b shares on Dec 31st, not 1.186b, with a forced sale of 80m shares in Q1 2025 (Jan 23, to be precise, taking them down to 1.186b) and another 8m in Q2, sor the number is now 1.178b. Does your table include the $46m gain on the sale of those 88m shares ? The gain is smaller than I would have thought, given the impressive share price gains of Eurobank recently, but it is because of the convoluted history of Fairfax's investment in Eurobank in 2014 and 2015, with a very strange weighted average cost of 2.20 euros. They first bought shares in 2014 at E0.31, and as there was a 100:1 split later on, this actually means their initial investment was at E31 !. But then they doubled down massively in 2015 (something some investors say you should never do, but never say never, right? So they bought way more shares in 2015 at E0.01 (split adjusted E1.00). This gave them a weighted average of E0.022, or E2.20 split adjusted. So the sales in 2025 at about E2.29 represent a surprisingly small gain. Going forward, the company will likely still have to sell small stakes to keep under 33.3%, but the gain will be far larger. For instance, at the current price of E3.44, the gain per share would be about E1.24 per share instead of E0.09. So all in all, they made a biggish investment in 2014 (E400m, at split-adjusted E31/share), then took swhat they describe as "one of our largest unrealized losses ever" in 2015, when they participated in a refinancing at split-adjusted E1.00 per share. So this investment represents a huge strikeout on the 2014 investment (still way underwater) and then an even huger homerun from the 2015 investment, up from E1.00 to E3.44.

I thought the Deep South would be anything between Hamilton and Pointe Pelée...

My thinking is different. If I have $5m in an RRSP, yes, I will have to pay half of that (actually, more like 54%) in taxes. But if I so I really have $2.3m in that account, and the federal government has $2.7m. Like for any other investment, I will have to pay 54% of income generated by my $2.3m in funds, and also 54% of whatever income is generated on the $2.7m of government assets that I am allowed to manage. But the amazing feature of an RRSP is that the government will allow me to keep the other 46% of income I might generate on ITS FUNDS. That $2.7m acts sort of like insurance float, investing other people's money with the return coming to you.

This is the part of your analysis that I am having trouble following. If Fairfax makes $4.5b in earnings, and distributes $300m in dividends, that's a lot of value being added every year. Is your model assuming some return on that extra $4.2b in retained earnings, or are you just extrapolating growth in earnings based on previous years' growth, under the assumption that a lot of that growth was also from retained earnings6

Aside? Is this a faulty transcription of Poseidon, which I would think would be the second biggest investment? Yes, that seems to be the case. In this day of AI-generated transcriptions, we have to be conscious of how these things sound, to know what some of the words are probably supposed to be (in bold): Net gains on investments in the quarter were again very healthy at $952,000,000 All in, our book value per share increased to $11.58 [obviously $1158] in the second quarter, up 10.8% in the first half of the year, adjusted for our $15 dividend. Our insurance and reinsurance companies are in great shape, writing over $33,000,000,000 of annualized premium worldwide. We benefit greatly from our scale and diversification and the exceptional talent and experience of our long serving presidents and the teams that run our insurance and reinsurance operations. I will now give you some additional detail on the components of our net earnings for the quarter. Our consolidated investment return was solid with a return of 2.6%, driven by increased interest and dividend income, strong net gains on investments, partially offset by the lower profits of associates. Consolidated interest and dividend income of $666,000,000 was up 8.5% year over year, benefiting from a growing investment portfolio and increased dividend income in the quarter. Profits of associates was $131,000,000 down by $90,000,000 compared to the 2024. Profits of associates continues to be driven by Eurobank and Poseidon Corp [not Aside, as in other transcriptions], offset this quarter by losses on the WaterUS [obviously Waterous] fund from mark to market unrealized losses in its portfolio.

Wow, yes, at -$903m, that's a big expense, in the context of $2304m in net earnings for the first half. Could you or one of the other people better versed in IFRS accounting provide some explanation of what the idea behind this charge is and how it applies to Fairfax in a different way from other insurers?

Yes, mea culpa. Eurobank as an associate (20-50% ownership) is equity accounted, meaning only Fairfax’s 33% proportion of Eurobank’s earnings (about E314m in Q1, not announced for Q2) get counted towards Fairfax’s Q2 earnings, so maybe about E105m pre-tax if Q2 is like Q1, nowhere near the E540m gain in market value of their stake.I should have just gone back to Viking’s table of equity holdings, where they are clearly divided according to accounting treatment (0-20% minority interest, 20-50% equity accounting, 50+% consolidated.) They still probably get over $60 per share in Q2 all things considered, with things like Orla and some of the Indian holdings and even Blackberry (all the equity holdings were pretty low on March 31st, and the non-US ones will have been boosted by US$ weakness), but Viking will probably blow us away with all the details.

I want to revise my pick ($50-$59) upwards by a couple of notches ($70-$79), even if I can't change my vote in the poll. Here's my thinking: Eurobank: shares up E0.45, so for 1.2b shares of Eurobank, divided by 21.4m shares of FFH o/s, that means $25 per FFH share, say $20 after tax; TRS: shares up $459 (hard to believe!), so 1.76m TRS for 21.4m shares means 459*1.76/21.4 = $37.7 per FFH share, say $30 after tax; Underwriting: 2023 and 2024 averaged $400m underwriting earnings per quarter, say $320m after tax. We think this was probably a particularly good quarter, so maybe $500m, so that would be 500/21.4=$23 per share Of course there are many more elements to consider, but most of them will be positive, and just the above 3 total to $73 per share. Maybe I should have gone for $80-89. Forex affects book value and fluctuates from quarter to quarter, no doubt positively this quarter (because of marked US dollar weakness) but this I believe does not affect earnings.

Pretty amazing that analysts are projecting $40-$50 in quarterly earnings for each of the next 4 quarters, mid-point $177.25 for the 4 quarters combined, and so at the current share price of $1788 this represents almost exactly 10x forward earnings, in a market where the S&P 500 P/E for forward earnings for the same periods is 25. And I believe that is with undiluted EPS (https://ycharts.com/indicators/sp_500_pe_ratio_forward_estimate), so the ratio with diluted EPS for the S&P would be higher still. I don't have a model, but based on the fact that recent quarters have had earnings (diluted EPS) between $37 and $50, despite currency headwinds, and this quarter we have currency tailwinds, no known major catastrophes, and big gains from the minority mark to market equity holdings, I am going to guess we are at $50 to $55 diluted EPS for Q2.

Agree with both points, but there are many other good reasons to own Fairfax. I don't feel I am part of the 'stock split crowd', and I agree that the high share price does put Fairfax in some pretty good company, but I can easily live without this feature.

Why? I agree it would be largely pointless, but what harm would it do to split it 25:1 and have a share price of about C$100 No need for another split for another 10 years or so, even with 25% annual share price gains, making us all financially independent? (Except maybe you, because you don't like them doing something that at least 99% of other companies do?)

Viking's used 1186m as the share count number, which would indeed be 1266m (from the AR) minus 80m from the January sale. It is hard to understand why Eurobank would have said in February that Fairfax owned 1209.2m shares after that sale, especially since we can calculate that by March 31st, Fairfax was down to 1183.6m shares. Bottom line is, we don't really know exactly how many shares they have, but it is probably less than 1183.6m now. For instance, we know that Eurobank repurchased 3.8m shares in June, suggesting that Fairfax must have sold another 1.3m or so. So a good guess is that we are at about 1182m shares now. We will get more news when Fairfax reports on Q2 in about a month. If Eurobank then also reports a higher number, perhaps we can infer that they are counting differently - for instance, maybe Eurobank is counting shares controlled by Watsa but not by Fairfax? Who knows.

Eurobank is on quite the tear. Not only is its share price up from $2.23 to 2.97 since the beginning of 2025, the EUR:USD exchange rate has kicked in another 10%, up from 1.04 to 1.18. Not to mention the Euro 10c dividend. Combining the 3 effects, Fairfax's biggest share investment (equity accounting) is up 56%, at least for the roughly 33% stake they have held on to. They had to sell 80m shares in January, to keep their stake under 33% as required regulators in Greece and Bulgaria, subsequent to which they owned 1209m shares, or 32.89%. Since then, Eurobank has announced its intention to repurchase up to 10% of its shares, and mentioned in May that it had been informed by Fairfax of Fairfax's intention to 'participate' in this share buyback programme, whatever that is supposed to mean. It may mean that they have sold some of their shares to Eurobank, to keep under the 33% limit. Obviously, to keep the same percentage stake, this would mean that Fairfax would need to sell up to 10% of its shares (about 121m), so the current number of shares owned may be less than 1209m. Indeed, as of Mar 31st, Fairfax said in their Q2 report that the fair value of its stake was $3160.6m, suggesting that they owned 1183.6 shares at that point, i.e. that they had sold about 25m more shares since the January sale. And the number may well have changed (i.e. decreased further) since then.