berkshire - cheap?

[SlowAppreciation]

This. This piece by SlowAppreciation, combined with the two Semper Augustus Client Letters about intrinsic value of BRK was the documents that gave me conviction to continue to add to BRK going forward.

 

Very much appreciated, SlowAppreciation, thank you for sharing. Damn good work. Please give your GF a gentle hug from me, it appears from your blog, that she is giving you a hand on your work.

Just read slow appreciation's piece. It's pretty well written. One big flaw is that in the valuation it ignores the float. While the value of that liability is less then book, it is definitely far from zero.

 

There's more than one way to skin a cat. I know one can do a float-based valuation of BRK, but felt it was the least-approachable from a casual reader perspective and also the one I'm least confident in doing accurately. So I left it out. I don't think it changes things all that much, and there's nothing wrong with baking in further conservatism to one's model.

[rb]

This. This piece by SlowAppreciation, combined with the two Semper Augustus Client Letters about intrinsic value of BRK was the documents that gave me conviction to continue to add to BRK going forward.

 

Very much appreciated, SlowAppreciation, thank you for sharing. Damn good work. Please give your GF a gentle hug from me, it appears from your blog, that she is giving you a hand on your work.

Just read slow appreciation's piece. It's pretty well written. One big flaw is that in the valuation it ignores the float. While the value of that liability is less then book, it is definitely far from zero.

 

There's more than one way to skin a cat. I know one can do a float-based valuation of BRK, but felt it was the least-approachable from a casual reader perspective and also the one I'm least confident in doing accurately. So I left it out. I don't think it changes things all that much, and there's nothing wrong with baking in further conservatism to one's model.

Well you count the securities per share but ignore the associated liability that finances those assets. Under this method every insurance company, indeed every financial institution on the planet is grossly undervalued.

[SlowAppreciation]

This. This piece by SlowAppreciation, combined with the two Semper Augustus Client Letters about intrinsic value of BRK was the documents that gave me conviction to continue to add to BRK going forward.

 

Very much appreciated, SlowAppreciation, thank you for sharing. Damn good work. Please give your GF a gentle hug from me, it appears from your blog, that she is giving you a hand on your work.

Just read slow appreciation's piece. It's pretty well written. One big flaw is that in the valuation it ignores the float. While the value of that liability is less then book, it is definitely far from zero.

 

There's more than one way to skin a cat. I know one can do a float-based valuation of BRK, but felt it was the least-approachable from a casual reader perspective and also the one I'm least confident in doing accurately. So I left it out. I don't think it changes things all that much, and there's nothing wrong with baking in further conservatism to one's model.

Well you count the securities per share but ignore the associated liability that finances those assets. Under this method every insurance company, indeed every financial institution on the planet is grossly undervalued.

 

Sorry, I misunderstood your original post.

 

This is a fair point, though I don't recall Buffett subtracting the liabilities when using the "two-column" valuation method. Curious why that might be.

[rb]

I think that maybe when Buffett does it, it's more of a rough sketch. He has no intention of spoon feeding a valuation of BRK. But I'm also pretty sure that when he looks to acquire an insurance company he doesn't count the float liability as zero.

 

Basically the float has to be discounted in some way. It's less then book but still a significant number.

 

Also in reference to your write-up. At a meeting in years past he did mention in passing that the basket of businesses BRK owns are probably worth about 14x pre tax.

[Guest longinvestor]

I think that maybe when Buffett does it, it's more of a rough sketch. He has no intention of spoon feeding a valuation of BRK. But I'm also pretty sure that when he looks to acquire an insurance company he doesn't count the float liability as zero.

 

Basically the float has to be discounted in some way. It's less then book but still a significant number.

 

Also in reference to your write-up. At a meeting in years past he did mention in passing that the basket of businesses BRK owns are probably worth about 14x pre tax.

 

Do you remember how long ago that was? Reason I ask is the basket has changed significantly since 2009. Was that statement made pre-2009?

[cubsfan]

Do you remember how long ago that was? Reason I ask is the basket has changed significantly since 2009. Was that statement made pre-2009?

 

I'm pretty sure it was 2012 meeting.

[rb]

It would have been in the 2011-2014 period.

 

As with everything WB says he says keep it in perspective. 14x pre tax is a very generic multiple. If you asked me what a US based company is worth without telling me the company I'd say about 14x pre tax. Let me demonstrate:

 

Take a hurdle rate of 9% (4% for risk free + 5% risk premium), US GDP growth will be about 4%, assume earnings growth will match GDP, and say a tax rate of about 30%. Throw that into a DCF and TADA!: you get a valuation of about 14x pre tax.

[LC]

Well you count the securities per share but ignore the associated liability that finances those assets. Under this method every insurance company, indeed every financial institution on the planet is grossly undervalued.

Isn't Berkshire generally posting underwriting profits, though? Their float is being paid for because they are better underwriters than other institutions. To other firms the float is a liability but for Brk it is an asset, or at least that is how I understand it.

[rb]

No LC. You could capitalize the underwriting profits in the valuation, which Slow Appreciation did. But that comes nowhere near to cancelling the liability.

 

Let me put it this way. As sure as the sun will come up tomorrow, a tiny part of that float is a payment that GEICO has to make to a body shop in the next week to pay for work that was done on John's car, because John got into an accident 2 weeks ago. Is that payment an asset or a liability to Berkshire? The float is made up of thousands examples like this spread out over time.

 

Let me put it another way. Let's say that you have insurance company A which has a consistent total ratio of 100.01% and insurance company B which has a consistent total ratio of 99.99%. Does A have a float liability and B have a float asset, thus making B way more valuable than A? Of course not.

 

Berkshire's present value of float is less then book. But it is still a huge liability and should not be ignored when valuing Berkshire.

[Guest longinvestor]

No LC. You could capitalize the underwriting profits in the valuation, which Slow Appreciation did. But that comes nowhere near to cancelling the liability.

 

Let me put it this way. As sure as the sun will come up tomorrow, a tiny part of that float is a payment that GEICO has to make to a body shop in the next week to pay for work that was done on John's car, because John got into an accident 2 weeks ago. Is that payment an asset or a liability to Berkshire? The float is made up of thousands examples like this spread out over time.

 

Let me put it another way. Let's say that you have insurance company A which has a consistent total ratio of 100.01% and insurance company B which has a consistent total ratio of 99.99%. Does A have a float liability and B have a float asset, thus making B way more valuable than A? Of course not.

 

Berkshire's present value of float is less then book. But it is still a huge liability and should not be ignored when valuing Berkshire.

 

From the AR,

 

So how does our float affect intrinsic value? When Berkshire’s book value is calculated, the full amount

of our float is deducted as a liability, just as if we had to pay it out tomorrow and could not replenish it. But to

think of float as a typical liability is a major mistake. It should instead be viewed as a revolving fund. Daily, we

pay old claims and related expenses – a huge $27 billion to more than six million claimants in 2016 – and that

reduces float. Just as surely, we each day write new business that will soon generate its own claims, adding to

float.

If our revolving float is both costless and long-enduring, which I believe it will be, the true value of this

liability is dramatically less than the accounting liability. Owing $1 that in effect will never leave the premises –

because new business is almost certain to deliver a substitute – is worlds different from owing $1 that will go out

the door tomorrow and not be replaced. The two types of liabilities, however, are treated as equals under GAAP.

A partial offset to this overstated liability is a $15.5 billion “goodwill” asset that we incurred in buying

our insurance companies and that is included in our book-value figure. In very large part, this goodwill represents

the price we paid for the float-generating capabilities of our insurance operations. The cost of the goodwill,

however, has no bearing on its true value. For example, if an insurance company sustains large and prolonged

underwriting losses, any goodwill asset carried on the books should be deemed valueless, whatever its original

cost.

Fortunately, that does not describe Berkshire. Charlie and I believe the true economic value of our

insurance goodwill – what we would happily pay for float of similar quality were we to purchase an insurance

operation possessing it – to be far in excess of its historic carrying value. Indeed, almost the entire $15.5 billion

we carry for goodwill in our insurance business was already on our books in 2000 when float was $28 billion.

Yet we have subsequently increased our float by $64 billion, a gain that in no way is reflected in our book value.

This unrecorded asset is one reason – a huge reason – why we believe Berkshire’s intrinsic business value far

exceeds its book value.

[LC]

Rb: I don't understand your point (then again I am pretty dense...)

 

If we take your example of company A (underwriting losses) and B (underwriting gains) and simplify the problem:

 

Imagine they both write their first business today.

Let's remove all investments of the float, assume they only hold cash.

 

Let's go forward 1000 years of this behavior, and then look at the companies.

 

Company A will now be bankrupt due to underwriting losses. Company B will have printed money continuously.

 

If we go back to time zero, we see the underwriting performance of A to be a huge liability which has wiped out the company, but a huge asset to B.

[rb]

Rb: I don't understand your point (then again I am pretty dense...)

 

If we take your example of company A (underwriting losses) and B (underwriting gains) and simplify the problem:

 

Imagine they both write their first business today.

Let's remove all investments of the float, assume they only hold cash.

 

Let's go forward 1000 years of this behavior, and then look at the companies.

 

Company A will now be bankrupt due to underwriting losses. Company B will have printed money continuously.

 

If we go back to time zero, we see the underwriting performance of A to be a huge liability which has wiped out the company, but a huge asset to B.

Yes I think you're misunderstanding what I'm trying to say. First of all no insurance company would just keep their float in cash. Secondly, I think there's some confusion around the words asset and liability where you're thinking more of a figurative sense of the words and I'm saying more of the financial sense of the words.

 

Let me clarify my A and B example a bit more by adding a couple more assumptions. 1: Both companies are behaving like insurance companies and the float is invested in a diversified portfolio of bonds and stocks. 2: Lets assume that both companies have floats of 100 billion.

 

I know my example is a bit extreme but it's so for a reason. Basically what i was trying to get with my example is that the financial performance of the two companies is essentially the same. Because one has a perpetual underwriting profit makes it a bit more valuable than the other which has a perpetual underwriting loss. But perpetual underwriting profits do not vanish a float liability which in my (updated) example would mean that one company is $100B more valuable than the other. Obviously incorrect.

 

Also if that were true then a hot stock tip: Progressive has had an underwriting profit for at least 10 years. Take out float liability and it's trading at 1/2 book. Happy days!

 

Basically, the quote from WB in the AR is spot on. All that means as I've been saying is that float is less then book but still a meaningful liability for BRK. I actually like this discussion because I've been thinking and struggling for a while to get to a good model to discount BRKs float. I hope this thread will help me get closer.

 

Final thought. If float was an asset rather then a liability then it means it has financial value. In turn that means the someone is willing to pay BRK to have those obligations taken off their hands. Obviously not true. There's nobody there, and WB would agree to that deal in a second. In fact large chunks of BRK float are made up of the obligations of other insurance companies that BRK took over. However those companies paid BRK (handsomely?) to do that.

[Jurgis]

Great post, rb.

[LC]

I've got a couple of thoughts but it's a bit late here so I'll put one here and mull over the rest for tomorrow. The reason I took out the investing portion when comparing 2 insurance companies is this:

 

Imagine an insurance company with negative underwriting profits. They are essentially a leveraged investment company. They pay X% for financial obligations, and try to invest to make X+Y%. The only value they add is via investing.

 

Take away their investing arm, and they will eventually drown. So their only real asset is their ability to invest profitably.

 

So why write insurance business? To grow AUM. So we can think of the liability associated with float as the cost for an investment manager to increase their AUM.

 

So if we look at it from that perspective, let's go back to underwriting profitably. Imagine investing with a fund manager. You pay him a % to invest with him. At some point you will redeem your investment, but when is that a REAL liability to his business? I can think of 3 scenarios:

-your clients will all jump ship in a crisis

-your performance is consistently just plain bad

 

and most importantly,

-your cannot replace any leaving AUM

 

Maybe I'm rambling but that's a thought I had.

[Guest longinvestor]

I'm no accountant.

 

Can someone build a simple model for what Buffett calls "dramatically lower liability to GAAP" replacing the word "dramatic" with ratios between 0.1 to 0.9 of GAAP reported liability and offset the goodwill $15.5B and the "float growth capability acquired using that goodwill resulting in $64 Billion of additional float since 2000 that in no way is reflected in the book" and what is the range of additional IV we are talking about.

 

I'm getting ridiculous numbers, enough to keep me awake!

 

 

[rb]

I'm no accountant.

 

Can someone build a simple model for what Buffett calls "dramatically lower liability to GAAP" replacing the word "dramatic" with ratios between 0.1 to 0.9 of GAAP reported liability and offset the goodwill $15.5B and the "$64 Billion of float growth capability acquired using that goodwill since 2000 which is in no way captured in the book" and what is the range of additional IV we are talking about.

 

I'm getting ridiculous numbers, enough to keep me awake!

Well that's the million dollar question isn't it? Btw, I don't think accounting has anything to do with it.

 

Let me propose another thought experiment. Because i think the underwriting profit murks the waters a lot. Let's put BRK in an underwriting loss. So I'll give BRK a 2 billion underwriting loss as opposed to 2 billion profit. Through my newfound powers I'll also increase the return of the securities portfolio by 4 billion a year. Ignore the tax effects and all other peculiarities.

 

In this case basically the profitability of BRK doesn't change. So it's value doesn't not change. Should make analysis easier though.

[LC]

It's value may change. All earnings are not created equal. If you think their underwriting earnings are more or less valuable than their investment portfolio earnings, then I would think their value would change.

[Guest longinvestor]

I'm no accountant.

 

Can someone build a simple model for what Buffett calls "dramatically lower liability to GAAP" replacing the word "dramatic" with ratios between 0.1 to 0.9 of GAAP reported liability and offset the goodwill $15.5B and the "$64 Billion of float growth capability acquired using that goodwill since 2000 which is in no way captured in the book" and what is the range of additional IV we are talking about.

 

I'm getting ridiculous numbers, enough to keep me awake!

Well that's the million dollar question isn't it? Btw, I don't think accounting has anything to do with it.

 

Let me propose another thought experiment. Because i think the underwriting profit murks the waters a lot. Let's put BRK in an underwriting loss. So I'll give BRK a 2 billion underwriting loss as opposed to 2 billion profit. Through my newfound powers I'll also increase the return of the securities portfolio by 4 billion a year. Ignore the tax effects and all other peculiarities.

 

In this case basically the profitability of BRK doesn't change. So it's value doesn't not change. Should make analysis easier though.

 

BTW, Buffett did not say that the float was an asset. It is just a dramatically lower liability than GAAP suggests.

 

Also, Goodwill of $15.5B came about in 2000 because of acquisition of GenRe (I think)? Also, my understanding is that many of Jain's policies written end up being costless and claimless!  Like the March Madness, Hole in one's etc. You collect, wait out the event, no? If there is ever a payout it gets paid over a very long time. Think Lloyds. I believe Buffett uses the term "long enduring" in this sense. 

 

[thelads]

Apologies in advance, as what I am about to propose may well be wrong. But I think of this simply as a choice of liability. It has debt like qualities, and is a source of financing. The point I think Buffett is making is that this is a long lived liability. Like a revolving credit facility, but one that is drawn almost all the time and whose size stays constant or grows. The alternative source of financing could be debt or equity, but both are higher cost. I have no idea what the right average life or duration of this liability should be, though if we look back to 1995, the float has only grown at GEICO, so you can say its at least 20 years. Probably a lot longer, but that depends on your view of automous driving, etc, other things affecting float. But just for arguments sake, lets say the right proxy is a 30 year bond as a replacement for float. What yield would need to be paid on that? Not just today, but in a normal environment? Even if we took a conservative price of say 2.5%, with a 30 year, zero coupon bullet bond, the fair value of such float (if labeled as a bond instead) would be ~50 cents on the dollar.

 

I may be wrong, but I think this is roughly the calculation he is doing, though what tenor as a proxy he uses, and what cost of debt, I don't know. But this also matches his comment that the float value is lower in a ZIRP world. All very intuitive I think also, but with this framework you can at least put a range of value on things. You can build a quick bond calculator to test this yourselves. It is obviously very sensitive to the interest rate you use. Obviously there is substantial value to growth in the float too.

 

The other thing that is really nice about float as a form of financing is that it is uncorrelated to other sources of financing, is non recourse, and doesn't have covenants or knock outs associated (except in a big tail event) - all of which allows WEB to operate counter cyclically even more so than he would otherwise.

[John Hjorth]

Great discussion - thank you all,

 

- - - o 0 o - - -

 

I think I read somewhere in the past, that Bershire lags analyst coverage... - not! It's actually one of the best covered companies in the world! - due to this board! That's meant as a compliment to my fellow board members!

 

- - - o 0 o - - -

 

Personally, I have a habit of thinking of the float and the deferred taxes in the Berkshire balance sheet as financial obligations basically free of interest [without discounting them], however the nature of those obligations separately are very different, and and the same time a bit similar.

 

I have never moved myself into the for me mentally unknown territory to try to estimate their net present value. If I should try to do it with the float, I would most likely use the provision triangles that we have access to from the 10-Ks, and try to discount the payouts using a WACC.

 

This approach is basically a calculation based on a liquidation perspective of the liability, based on hindsight [histocal experience] projected to the future, the provision triangles are naturally dynamic over time etc.etc.

 

After all, what rb is pursuing, is a method for doing a rough estimate of the overall discount to apply to the Berkshire float, so I think - with all the flaws inherent in this method - that it's better to try that than just doing nothing, if you're pursuing to do the discount.

 

Edit:

 

The percentage table on lower part of p. 91 in the 2016 10-K will be where to start, and then decompose total float on GEICO, Gen Re, BHRG & BH Primary to do the math.

 

Edit 2:

 

Total float year end 2016 is stated to USD 91.6 B, ref. the February 25 2017 Press Release. At p. 87 at the top in the 2016 10-K  there is a specification of the part of the float, totalling USD 76.918 B, ref. that separate post in the Berkshire balance sheet year end 2016.

[vinod1]

“[if] I were offered $7 billion for [$7 billion of] float

and did not have to pay tax on the gain, but would

thereafter have to stay out of the insurance business

forever—a perpetual noncompete in any kind

of insurance—would I accept that? The answer is

no. That’s not because I’d rather have $7 billion of

float than have $7 billion of free money. It’s

because I expect the $7 billion to grow.”

 

—Warren Buffett, 1996 Annual Shareholders’

Meeting, as quoted in Outstanding Investor Digest.

 

“If you could see our float for the next 20 years and

you could make an estimate as to the amount and

the cost of it, and you took the difference between

its cost and the returns available on governments,

you could discount it back to a net present value.”

 

—Warren Buffett, 1992 Annual Shareholders’

Meeting, as quoted in Outstanding Investor Digest.

 

I wrote up BRK in 2010, here is the extract related to the float valuation:

 

The key question concerns the value of float. Float is money an insurance company holds but does not own. Float arises because premiums are received before losses are paid and this interval can be of several years. During this interval, the insurer can invest the money for its own gain. The premiums are often not sufficient to cover all the costs and insurance company has to part with some of the returns generated out of float to make up for the shortfall.

 

To get an understanding of the theoretical basis for assigning value to float, let us take an example. Assume you are offered the following proposition: You are offered $100 to be paid back after the end of 50 years. You do not have to pay any interest and also get to keep all the money earned in the interim from investing the $100. The only restriction is that you can only invest in high quality investment grade bonds and treasuries.

 

How much would a smart business man be willing to pay for such an offer? If you can estimate the market rates of interest for long term bonds are going to be about 4% then the business man should be willing to pay about $86. (The business man can invest $14 at 4% for the 50 year period to end up with $100.)

 

Of course, the business man would not go through the trouble of investing to end up with no profit so he is going to offer something a little lower to make relatively low risk profit.

 

Insurance float for Berkshire is a lot like this example with few additional benefits. First, the $100 amount is likely to grow around the nominal GDP growth rate of 4-5%. Second, as long as the business is not wound up there is no need to pay back the $100. The effective length of the investment is longer than 50 years. Third, there is a strong possibility of getting paid something like 1-2% for the privilege of holding the money.

 

So what would a business man offered this proposition pay for this? It should be obvious that it would be at least $100 without even performing any complicated math.

 

A conservative value of this float can be calculated mathematically with a few assumptions. Assuming no growth in float in the future; that float earns at the treasury bond rate (4%); discounted at the treasury bond rate; and that the insurance segment would not liquidated in the foreseeable future; the present value of the float is simply the amount of float i.e. $100 using the constant growth dividend model.

 

There are two caveats to this calculation of the value of float.

 

(1) Shareholder incurs an additional cost for the float through an insurer due to tax penalty. This cost has been estimated at around 1% by Buffett.

 

(2) There is uncertainty in the true cost of float as it a near certainty that there would be periods of underwriting losses. Both these complications however should not alter the final value in the above calculation as it would be possible to conservatively earn enough above the Treasury bond rate on the float to mitigate this additional cost.

 

Vinod

 

 

[racemize]

The issue here is that this would add one or two hundred billion to berkshire IV, but the market simply doesn't value Berkshire or any insurance this way. Or saying it another way, this boost in value now would likely come at an extreme cost in value growth in the future.

[vinod1]

To put it more concisely, if you are given $100 today and every year going forward you would be given another $3 to $4 for the next 100 years with the caveat that you might also have to give back $3 to $5 once every 10 years (to simulate underwriting loss), what would you pay for that privilege?

 

I would pay $100 for that. Note that, you would get that $100 back immediately to be paid back in 100 years.

 

Vinod

 

[vinod1]

The issue here is that this would add one or two hundred billion to berkshire IV, but the market simply doesn't value Berkshire or any insurance this way. Or saying it another way, this boost in value now would likely come at an extreme cost in value growth in the future.

 

This would only add about $90 billion in float value which the float amount at end of 2016. Part of this float value shows up in the goodwill so you have to account for that. Part is always held in cash equivalents always. This part would not have full face value. If you make these adjustments, I think the market is valuing BRK in this manner, at least implicitly from the P/BV multiple.

 

When you estimate value of BRK from various methods they tend to cluster together closely. So float based valuation does not radically increase BRK valuation.

 

Vinod

[Guest longinvestor]

The issue here is that this would add one or two hundred billion to berkshire IV, but the market simply doesn't value Berkshire or any insurance this way. Or saying it another way, this boost in value now would likely come at an extreme cost in value growth in the future.

 

This would only add about $90 billion in float value which the float amount at end of 2016. Part of this float value shows up in the goodwill so you have to account for that. Part is always held in cash equivalents always. This part would not have full face value. If you make these adjustments, I think the market is valuing BRK in this manner, at least implicitly from the P/BV multiple.

 

When you estimate value of BRK from various methods they tend to cluster together closely. So float based valuation does not radically increase BRK valuation.

 

Vinod

 

Which brings us full circle to the theme of this thread. All of the methods used to value BRK , as elegant and/or conservative have resulted in Mr. Market being wrong to grossly wrong about BRK's value. The grossest error is the use of the near universal "let's put a 10x multiple on the Pre-tax earnings". Semperaugustus comes closer than most in properly valuing Berkshire. Being elegant in paper calculations doesn't preclude being ridiculously wrong. In a world where ridiculousness is rampant to unhinged on the other side. Think FANG.

 

Berkshire's float is invested neither in bonds nor securties but increasingly in long lived assets, whose earnings in turn are invested in other businesses (the real estate within BHE as an example), plus arguably their own kind of deferred tax float. Buffett calls this rabbits making more rabbits.

 

It all adds up to the 20% earnings growth we're seeing since 1999.

 

Thank heavens the market is wrong!