Why not do a straight-up valuation of Berkshire? While the exercise may not be for
first-timers, it shouldn't be all that tough if we keep it simple. Problem is, every time folks seem to take a shot at something like this, they take one too many short-cuts, use one too many proxies for value. We start seeing new directions in valuation not used anywhere else. Not that that’s necessarily bad – but maybe it’s not always necessary.
There are a couple of leaps of faith with this, but hopefully we'll explain adequately. The first thing we need to do as we get underway with this is strip out cash+investments and debt – which hopefully we agree can be valued 'dollar-for-dollar' – and then add them back after we figure the values of the underlying businesses, the 'enterprise values'.
While a discussion like this often revolves around PE’s (and I should say, I get easily confused when we see references to ‘PE’ kicking around that are actually on a pretax basis), sometimes it’s worth taking the leap to ‘enterprise value’ and ‘earnings before interest and taxes' - which can be helpful for comparisons to businesses with varying capitalizations. With EV/EBIT we can make apples-to-apples comparisons of underlying businesses with different capital structures -- valuing the underlying businesses, and then looking at the capitalization separately.
We can look at Berkshire’s non-insurance companies first - if we develop an estimate of this year’s earnings before interest and taxes – EBIT -for each business category delineated in the company reports, and then compare them to similar publicly traded companies’ Enterprise Value-to-EBIT multiples. Since we’re using some estimates we’ll try to make of Berkshire units’ 2011 EBIT, we’ll use consensus estimates of 2011 EBIT to get our ‘comparable company’ multiples.
It’s tempting at this point to ‘normalize’ our Berkshire EBIT estimates to some near-future ideal, but if we do, we also have to normalize the comparison companies. For example, if we think BNSF’s 2011 earnings are not representative, we’d have to ask if comparison-company’s Union Pacific's earnings are already at normalized levels. We wouldn’t want to adjust one without also normalizing the other.
The numbers on the schedules here are illustrative of the process, and not meant to be precise to a particular period. We’ve had enough material fluctuations in both prices (the principal component of EV) and the rush of recent Q2 earnings releases that their outdated almost as I write this. I also didn’t try to update the Berkshire investment holdings (Amex, Coke, etc) and I just kept with the Q2 reporting. Feel free to update them as you like.
For comparison companies, we have:
- Burlington Northern==>Union Pacific, Norfolk Southern; CSX
- MidAmerican==> Dte; Duke; Xcel
- Finance (Clayton, Cort,XTRA, etc)==> McGrath; Aaron’s
- Marmon==> Actuant; Carlisle; Tyco
- McLane==>SYSCO; Loblaw
- Other Mfg (Iscar, Mansville, Shaw, Fruit/Loom, etc)==>60 midcap industrial machining and bldg material companies, with EBIT % to revenues of 8% to 20%
- Other Services (NetJets, BusWire, TTI, DQ, etc)==> 8 midcap business service companies with EBIT to revenue percentages of 8% to 20%.
- Retail==> 40 midcap retailers with EBIT % to revenues of 5% to 15%
By all means, pick your own Berkshire comparisons use your own preferences.
I tended towards the higher end of the comparison companies for application to the Berkshire units – but again use your own judgment in selecting multiples. As I noted, the market has been bouncing around, and earnings releases have been coming out this week, so the actual numbers shown here already need updating. The process is what matters, though.
The peers generally had pretty close valuations within their groups – no crazy disparities—though as I noted, I tended to use the higher-multiple comparisons for Berkshire values. With this, we have values of the businesses before cash or debt.
2011 estimates ($millions)
Next up, the insurance subs – but maybe we can look at them ‘straight-up’ with normalized capitalization. Cash/securities are integral to the insurance subs operations, so we can’t dismiss that component altogether, but Berkshire’s units are overcapitalized versus their peers – a result of successful float deployment in the investment portfolio, etc, etc. We’ll look at the overall cash and securities separately, but for valuation purposes we’ll earmark a ‘comp’ level for the businesses.
This segment is interesting, and no doubt a bit controversial. Buffett implied in his ‘two column’ approach that we ignore insurance company valuations and look at just the cash+investment holdings. Since he came out with that approach, some have lamented that this understated Berkshire’s values by the unique value of the insurance subs, with their low-cost and well-deployed float. While there might be some complaint there, from a more straight-forward valuation perspective, Buffett’s approach with his two-column view has a particular elegance that I for one didn't appreciate for quite awhile.
If we look around the insurance industry, we see valuations that are particularly close to book value – in fact, these days, most insurers have valuations that are under book. The two other global re-insurers, Swiss Re and Munich Re, are about 80% of book. Travelers, Prudential, CNA, are all well under. Of the majors, Progressive, Aflac, and Chubb (just barely) are over book, but none by crazy multiples (like 2x book or anything).
So getting back to ‘two-column,’ if we did carve out and value the subs independently of the rest of the ‘investment column’, we’d also want to carve out some net assets.
If for example we set up a separate ‘insurance sub’ column and assigned a market value to it, we’d have to move some assets from the investment column to support that now-valued business. If for argument, we valued the Berkshire Insurance subs ‘at book’, we’d add market value to the insurance sub column, but at the same time back out the equivalent amount of securities from the original investment column. If we valued a Berkshire insurance sub (on a prevailing peer basis) at less than book, we’d transfer over more of those 'NBV' cash and securities than we added in business value. Vice versa if we assigned a Progressive- or Aflac-level value).
Buffett cleanly cut through this by implicitly assuming’ book’ – certainly a higher valuation than the market affords most insurers – but not up there in 'enhanced value' Progressive/Aflac territory. By picking up the full value of the investment column, he had already picked up the benefit of the ‘float’, etc.
We’ll go out on a limb and take a flyer at putting a value on the ‘normalized’ insurance subs. We’ll remember that when Ajit Jain gave his interview India, he said he sent the money his unit made to Warren's ‘black box’ in Omaha for deployment. We’ll value the insurer first, and what was sent to the black box separately.
For these purposes, we’ll see that the actual total value we assign the insurance subs is not necessarily as consequential to our overall Berkshire valuation as the market multiple (or discount) to book, which is the pivotal ingredient.
Stepping out a bit – and this is the most subjective element of the valuation (and again, for illustration purposes) – we’ll peg GenRe at about a 25% market-multiple-premium to its major competition, SwissRe and MunichRe (conveniently all the way up to 1x book); value Geico about with the industry’s highest-multiple Progressive, which many would consider reasonable; give Ajit a 50% premium to the industry (we can crank that up if we like), and put the remaining subs right up there at 1x book.
Feel free to pick your own values - this is an area where you can drill down to qualitative differences between the companies. We should keep in mind that the median market valuation for major insurers these days is 0.75x book, though, so we should think through the differentators.
Again, we’re talking insurance operations, not the black box in Omaha, and simply a refinement from 'two-column' for those unsatisfied with Buffett's implied appraoch. Pick your own valuation numbers.
Finally, we pull in the loose ends and bring it all together.
The cash/securities values are from the latest Q – not adjusted for subsequent excitement (feel free). We then are taking out the cash, etc, that we used to support the straight-up insurance valuation - to get them to the hypothetical 'comp'.
When we started this process we segregated out debt – from both the Berkshire subs and the comps – and now we have to re-introduce the Berkshire debt.
There will eventually be some tax liability attached to the gains in the portfolio, when they are realized. If we are inclined, we can wade into the ‘deferred taxes’ pit and attempt to put a present value on that perhaps-distant tab.
An optional note: if you have any subjective factors you want to include– plus or minus – say for your view on the future security and performance of that ‘black box’ in Omaha — go for it. I prefer sticking to what I reasonably quantify, but the market does apply discounts and premia for these things from time to time.
Putting it all together – and again using some place-holder assumptions that we’ll want to refine, we get a ‘built-up’ value of – again just in this illustration -- $144k per share. [I’m well past my 3-beer posting minimum, so if I’ve missed anything, obvious or not, feel free to throw it in].
The approach is straight-up --no major reliance on one-step-removed derivative proxies of value. And you don’t have to guess at all that much – just plug in the numbers -- and if you want more complexity layered into this, you can add it.
first-timers, it shouldn't be all that tough if we keep it simple. Problem is, every time folks seem to take a shot at something like this, they take one too many short-cuts, use one too many proxies for value. We start seeing new directions in valuation not used anywhere else. Not that that’s necessarily bad – but maybe it’s not always necessary.
There are a couple of leaps of faith with this, but hopefully we'll explain adequately. The first thing we need to do as we get underway with this is strip out cash+investments and debt – which hopefully we agree can be valued 'dollar-for-dollar' – and then add them back after we figure the values of the underlying businesses, the 'enterprise values'.
While a discussion like this often revolves around PE’s (and I should say, I get easily confused when we see references to ‘PE’ kicking around that are actually on a pretax basis), sometimes it’s worth taking the leap to ‘enterprise value’ and ‘earnings before interest and taxes' - which can be helpful for comparisons to businesses with varying capitalizations. With EV/EBIT we can make apples-to-apples comparisons of underlying businesses with different capital structures -- valuing the underlying businesses, and then looking at the capitalization separately.
We can look at Berkshire’s non-insurance companies first - if we develop an estimate of this year’s earnings before interest and taxes – EBIT -for each business category delineated in the company reports, and then compare them to similar publicly traded companies’ Enterprise Value-to-EBIT multiples. Since we’re using some estimates we’ll try to make of Berkshire units’ 2011 EBIT, we’ll use consensus estimates of 2011 EBIT to get our ‘comparable company’ multiples.
It’s tempting at this point to ‘normalize’ our Berkshire EBIT estimates to some near-future ideal, but if we do, we also have to normalize the comparison companies. For example, if we think BNSF’s 2011 earnings are not representative, we’d have to ask if comparison-company’s Union Pacific's earnings are already at normalized levels. We wouldn’t want to adjust one without also normalizing the other.
The numbers on the schedules here are illustrative of the process, and not meant to be precise to a particular period. We’ve had enough material fluctuations in both prices (the principal component of EV) and the rush of recent Q2 earnings releases that their outdated almost as I write this. I also didn’t try to update the Berkshire investment holdings (Amex, Coke, etc) and I just kept with the Q2 reporting. Feel free to update them as you like.
For comparison companies, we have:
- Burlington Northern==>Union Pacific, Norfolk Southern; CSX
- MidAmerican==> Dte; Duke; Xcel
- Finance (Clayton, Cort,XTRA, etc)==> McGrath; Aaron’s
- Marmon==> Actuant; Carlisle; Tyco
- McLane==>SYSCO; Loblaw
- Other Mfg (Iscar, Mansville, Shaw, Fruit/Loom, etc)==>60 midcap industrial machining and bldg material companies, with EBIT % to revenues of 8% to 20%
- Other Services (NetJets, BusWire, TTI, DQ, etc)==> 8 midcap business service companies with EBIT to revenue percentages of 8% to 20%.
- Retail==> 40 midcap retailers with EBIT % to revenues of 5% to 15%
By all means, pick your own Berkshire comparisons use your own preferences.
I tended towards the higher end of the comparison companies for application to the Berkshire units – but again use your own judgment in selecting multiples. As I noted, the market has been bouncing around, and earnings releases have been coming out this week, so the actual numbers shown here already need updating. The process is what matters, though.
The peers generally had pretty close valuations within their groups – no crazy disparities—though as I noted, I tended to use the higher-multiple comparisons for Berkshire values. With this, we have values of the businesses before cash or debt.
2011 estimates ($millions)
Non-insurance businesses - excluding cash & debt Business Group Pretax Interest EBIT EV/EBIT Ent.Value BNSF 4,200 750 4,950 9.0 44,550 MidAmer.(90%) 1,800 1,000 2,800 11.5 32,200 Finance 750 700 1,450 10.0 14,500 Marmon (80%) 750 750 8.5 6,375 McLane 400 400 9.0 3,600 OthMfg 2,200 2,200 9.2 20,240 OthSvce 1,000 1,000 10.0 10,000 Retail 220 220 8.0 1,760 Tot NonInsur. 11,320 2,300 13,770 9.5 133,225
Next up, the insurance subs – but maybe we can look at them ‘straight-up’ with normalized capitalization. Cash/securities are integral to the insurance subs operations, so we can’t dismiss that component altogether, but Berkshire’s units are overcapitalized versus their peers – a result of successful float deployment in the investment portfolio, etc, etc. We’ll look at the overall cash and securities separately, but for valuation purposes we’ll earmark a ‘comp’ level for the businesses.
This segment is interesting, and no doubt a bit controversial. Buffett implied in his ‘two column’ approach that we ignore insurance company valuations and look at just the cash+investment holdings. Since he came out with that approach, some have lamented that this understated Berkshire’s values by the unique value of the insurance subs, with their low-cost and well-deployed float. While there might be some complaint there, from a more straight-forward valuation perspective, Buffett’s approach with his two-column view has a particular elegance that I for one didn't appreciate for quite awhile.
If we look around the insurance industry, we see valuations that are particularly close to book value – in fact, these days, most insurers have valuations that are under book. The two other global re-insurers, Swiss Re and Munich Re, are about 80% of book. Travelers, Prudential, CNA, are all well under. Of the majors, Progressive, Aflac, and Chubb (just barely) are over book, but none by crazy multiples (like 2x book or anything).
So getting back to ‘two-column,’ if we did carve out and value the subs independently of the rest of the ‘investment column’, we’d also want to carve out some net assets.
If for example we set up a separate ‘insurance sub’ column and assigned a market value to it, we’d have to move some assets from the investment column to support that now-valued business. If for argument, we valued the Berkshire Insurance subs ‘at book’, we’d add market value to the insurance sub column, but at the same time back out the equivalent amount of securities from the original investment column. If we valued a Berkshire insurance sub (on a prevailing peer basis) at less than book, we’d transfer over more of those 'NBV' cash and securities than we added in business value. Vice versa if we assigned a Progressive- or Aflac-level value).
Buffett cleanly cut through this by implicitly assuming’ book’ – certainly a higher valuation than the market affords most insurers – but not up there in 'enhanced value' Progressive/Aflac territory. By picking up the full value of the investment column, he had already picked up the benefit of the ‘float’, etc.
We’ll go out on a limb and take a flyer at putting a value on the ‘normalized’ insurance subs. We’ll remember that when Ajit Jain gave his interview India, he said he sent the money his unit made to Warren's ‘black box’ in Omaha for deployment. We’ll value the insurer first, and what was sent to the black box separately.
For these purposes, we’ll see that the actual total value we assign the insurance subs is not necessarily as consequential to our overall Berkshire valuation as the market multiple (or discount) to book, which is the pivotal ingredient.
Stepping out a bit – and this is the most subjective element of the valuation (and again, for illustration purposes) – we’ll peg GenRe at about a 25% market-multiple-premium to its major competition, SwissRe and MunichRe (conveniently all the way up to 1x book); value Geico about with the industry’s highest-multiple Progressive, which many would consider reasonable; give Ajit a 50% premium to the industry (we can crank that up if we like), and put the remaining subs right up there at 1x book.
Feel free to pick your own values - this is an area where you can drill down to qualitative differences between the companies. We should keep in mind that the median market valuation for major insurers these days is 0.75x book, though, so we should think through the differentators.
Again, we’re talking insurance operations, not the black box in Omaha, and simply a refinement from 'two-column' for those unsatisfied with Buffett's implied appraoch. Pick your own valuation numbers.
At normalized capitalization Sub BkValue Multiple MktCap Geico 7,100 1.7 12,000 GenRe 18,000 1.0 18,000 BHRG 5,000 1.2 6,000 BH other 2,000 1.0 2,000 TotInsur 32,100 38,000 Net value pick-up vs book 5,900
Finally, we pull in the loose ends and bring it all together.
The cash/securities values are from the latest Q – not adjusted for subsequent excitement (feel free). We then are taking out the cash, etc, that we used to support the straight-up insurance valuation - to get them to the hypothetical 'comp'.
When we started this process we segregated out debt – from both the Berkshire subs and the comps – and now we have to re-introduce the Berkshire debt.
There will eventually be some tax liability attached to the gains in the portfolio, when they are realized. If we are inclined, we can wade into the ‘deferred taxes’ pit and attempt to put a present value on that perhaps-distant tab.
An optional note: if you have any subjective factors you want to include– plus or minus – say for your view on the future security and performance of that ‘black box’ in Omaha — go for it. I prefer sticking to what I reasonably quantify, but the market does apply discounts and premia for these things from time to time.
Putting it all together – and again using some place-holder assumptions that we’ll want to refine, we get a ‘built-up’ value of – again just in this illustration -- $144k per share. [I’m well past my 3-beer posting minimum, so if I’ve missed anything, obvious or not, feel free to throw it in].
Value components: Non-insurance enterprise value 133,225 from the first chart above Insurance mkt cap (x-excess invest) 38,000 from insurance sub chart Cash & investment(x-derivatives) 150,109 from the Q2 10Q, we hope +/- invest. change since end of qtr (?) update of the material securities holdings less - cash/inv. reclass to insur. nbv (32,100) from insurance sub chart Other invst, net (incl derivatives) 5,035 from the Q less - debt (56,717) from the Q +/- NPV of def'd tax liability (?) go for it +/- any subjective elements --- Total 237,551 A-share equivalents 1.650 Per share 143,970
The approach is straight-up --no major reliance on one-step-removed derivative proxies of value. And you don’t have to guess at all that much – just plug in the numbers -- and if you want more complexity layered into this, you can add it.
