Message Font: Serif | Sans-Serif
No. of Recommendations: 6
I still think that it is quite difficult to come up with an intrinsic value estimate for Berkshire under the $160,000 to $165,000 range unless one wishes to use extremely bearish projections for Berkshire's future growth and also demands a high projected return (which would require a high discount rate on top of bearish projections). For example, if one assumes that book value will only grow by 3% annually over the next decade, book value per A share might only be around $145K in 2022 and the P/B ratio would likely contract due to such poor results, maybe to around 1.0x book meaning A shares would trade at $145K in ten years. If a buyer today demanded a 10% expected return and bought into these bearish assumptions, one should not pay more than $56K/A share - that's the price you could pay today and still expect 10% annualized if the ending share price in ten years is $145K. That's about 50% of book value per share.

I think this scenario is ridiculous but it would represent a bearish outlook based on the inputs provided.

While I think it is possible to get to intrinsic value estimates in the $180K to $190K range based on still reasonable assumptions, I prefer to not get overly aggressive. This is why I use a 8x pre-tax multiple (around 12x after-tax) on the non-insurance subsidiaries even though Buffett has strongly implied that he uses a 10x multiple. And unlike Tilson's model, I do not assume any normalized underwriting profits at all ... which is conservative given Berkshire's history of underwriting profits over multiple cycles.

Using the float based model, one can get pretty much any valuation desired which is a huge weakness and small changes in variables such as the assumed rate of return on investments, assumed cost of float, growth of float, and discount rates can radically change the valuation. So while I still think the float model is perhaps the most intellectually appropriate way to measure Berkshire's value, it may also be the least useful AND the most prone to investor biases given the sensitivity of the valuation to input assumptions. I still use the float based model but cross check with other models and try to use conservative assumptions on the input variables.

I think tangible book is quite useless for Berkshire due to the obvious economic goodwill that exists in many of the non-insurance subsidiaries not to mention many of the primary insurance businesses as well as GEICO. GEICO's economic goodwill is huge if one considers Progressive to be a comparable company and looks at what the market assigns to Progressive's goodwill. I do track P/TBV but it isn't particularly meaningful.

In my opinion, having multiple models for Berkshire (as well as other companies) is very important because it allows for cross checking and also imposes a more rigorous test when it comes time to consider selling. It is important to not mess with input variables randomly or based on transitory factors ... but obviously sometimes changes are justified. It's just important to make sure that psychological factors are not prompting model changes. It's easy enough to fall into that trap especially with the float based model which is why I didn't recognize the shares as fully valued or overvalued in late 2007 (or late Sept/early Oct 2008) and failed to sell a meaningful number of shares.
Print the post  


When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.