I really enjoyed the first edition of this book， which I read almost 20 years ago。 The two approaches to valuation prescribed in that edition were refreshingly straightforward。 The most conservative approach was the asset-based valuation。 Estimate a company’s asset reproduction value and buy at a discount。 The next-most conservative approach was the earnings power-based valuation。 Calculate normalized earnings （assuming no growth） and divide by the cost of capital。 Both approaches were to-the-po I really enjoyed the first edition of this book， which I read almost 20 years ago。 The two approaches to valuation prescribed in that edition were refreshingly straightforward。 The most conservative approach was the asset-based valuation。 Estimate a company’s asset reproduction value and buy at a discount。 The next-most conservative approach was the earnings power-based valuation。 Calculate normalized earnings （assuming no growth） and divide by the cost of capital。 Both approaches were to-the-point and practically-applicable。I was therefore quite surprised when Bruce Greenwald recently stated in an interview that he thought the first edition of the book wasn’t very good， and that this second edition， 20 years later， was better。 The second edition would apparently incorporate a “returns-based” framework for analyzing growth stocks -- going far beyond the two methods described above。 Intrigued， I bought a copy。Long story short -- I think this second edition -- in particular Chapter 8， the new chapter on valuing growth stocks -- completely loses the plot。 In fact， I think that the “returns-based” model ruins the entire book。 At the very least， it serves to remind the reader that the （five！） authors have limited real-world investing experience （on any real scale）。 What follows is a （very） long review。 I’d usually refrain from writing a review at all， but given that this book is supposed to be one of the seminal texts in the field of value investing （and given that I am an avid value investor）， I felt that its shortcomings should be highlighted。Let me get straight to the point: no one who actually runs money for a living evaluates growth stocks in the manner prescribed in Chapter 8。 The authors pitch their “returns-based” approach as a superior alternative to the DCF， because， as they note， the DCF is sensitive to assumptions and results in a much greater variance of outcomes。 The authors contend that， as a result of these variances， calculating the intrinsic value of a rapidly growing company is “impossible”。 I find this to be ridiculous。 Any attempt at valuation will be subject to a significant degree of error。 The idea is not total precision。 The DCF is king because it is the literal definition of valuation: a company is worth the sum of its future free cash flows， discounted back to the present。 Full stop。 The earnings power value approach prescribed earlier in the book IS a DCF -- just one that assumes no growth！ Moreover， the earnings power value approach suffers from huge variances depending on the discount rate chosen。 Yet the authors are OK endorsing it。 Just because a rapidly growing company will result in a wider range of DCF outcomes doesn’t mean that you throw the whole methodology away！Sadly， the proposed replacement methodology is limited to the point of being useless。 The authors’ recommended process for analyzing growth stocks is as follows: take a company’s current distributable earnings yield， and add to it your estimate of that company’s long-run organic growth rate。 That’s it。 The combination of these two things is your expected return。 If this return is higher than your cost of capital， then invest in said growth stock！ If it is lower， then don’t。The authors then distinguish between “market-based” organic growth， and organic growth driven by “active investment”。 I have no qualms with this distinction， and the “active investment” piece speaks to the importance of return on capital exceeding cost of capital。But then there is a needlessly complex further digression highlighting the fact that the organic growth component of the total return equation above doesn’t actually equate to the “real” return that an investor should expect if a company’s market value is above its intrinsic value。 There is also an appendix to Chapter 8 that will only serve to confuse readers。 The second part of this appendix （or maybe the entire appendix） is written by someone who I'd bet has never managed real money。But now I’m digressing。 As I noted， the authors’ method of analyzing growth stocks with their new approach is extremely limited， for several reasons。 First， their model cannot handle long-run organic growth rates much higher than 10%， for otherwise the target company would eventually grow to be the size of the entire economy。 The authors point this out， but then do nothing about it。 So the “returns-based” model does not work for ANY company that is growing quickly （say， 15-25-35% a year for the next 5-7 years）。 The model also does not work for companies that are losing money or have no retained earnings， or for companies that have dramatically varying year-to-year growth rates。 I would say that the above three categories encompass perhaps 90% of all of the interesting growth stocks out there。 A DCF approach， of course， works for all three。The authors’ method instead works for only ONE type of growth stock opportunity: the slow-and-steady grower that is profitable。 The stalwart-type company。 But even for this type of company， the results of the model are nonsensical。 Consider a simple example。 Imagine a dominant， competitively-advantaged company that is 100% equity financed， that has no excess cash or debt， and that trades at a 20x P/E。 Imagine that this company is able to reinvest all of its earnings each year and generate a 20% return on investment。 Imagine that this company has a very long growth runway， and can keep reinvesting pretty much indefinitely。 In other words， this is a company that grows both revenue and earnings at 20% annualized for many years。 Imagine that your cost of capital is 10%。Based on the authors’ methodology， this company’s distributable earnings yield is 0%， given the fact that all of its earnings are being reinvested。 This 100% “active reinvestment”， at a 20% ROI versus our 10% cost of capital， implies a “value creation ratio” （authors’ term） of 2。0x。 2。0x multiplied by the 5% retained earnings yield results in an expected future growth “return” of 10%。 So， to summarize the authors’ approach: a 0% distributable earnings yield + a 10% growth return equals your total expected return of 10%。Given that our assumed cost of capital is also 10%， and given that we haven’t taken into account “fade rates” yet （another concept introduced by the authors with almost no real-world value）， we should be indifferent as to whether or not we make this investment。 After accounting for “fade”， the authors’ methodology would actually recommend avoiding this investment altogether！Any seasoned practitioner will tell you that a dominant company trading at 20x earnings that can reinvest all of its earnings at 20% incremental returns on capital for a very long time。。。 is very likely to be a good investment。 If you plug the same figures above into a 10-year DCF and assume a terminal growth rate of 5%， the DCF will tell you that this particular company is worth nearly 40x earnings， rather than 20x。 Yet with the authors’ model， the correct answer is to pass on the investment！ No sane investor would pass on this！The authors’ model would have dismissed Heico a decade ago as a pass。 The model would have dismissed Copart a decade ago。 The model would have dismissed Constellation Software a decade ago。 The model would have dismissed Old Dominion Freight Lines a decade ago。 Go and take a look at what all of those stocks have done since 2010。 The authors’ model would have dismissed any number of similarly great investments across the globe。Even the authors’ specifically-chosen examples in the book illustrate the uselessness of their “returns-based” model！ For example， the authors apply their model to Intel， and the model ends up suggesting that Intel stock offered a better prospective return in March of 1999 （at 10。6% annualized and 5。5% “adjusted”） than in March of 2003 （at 8。2% annualized and 3。7% “adjusted”）。 Zooming out for a moment， in March of 1999， the internet bubble was in full swing。 Intel was trading at a near-peak valuation of almost 40x earnings。 In March of 2003， the internet bubble had popped and Intel was trading at a much more reasonable mid-teens normalized earnings multiple。 Yet the authors （of a book on value investing， no less！） suggests that Intel would have offered higher prospective returns in 1999 than in 2003。What actually happened？ In the 10 years from March of 1999 onward， Intel returned -7。3% a year。 In the 10 years from March of 2003 onward， Intel returned +5。2% a year。 Intel was a much better buy at the bottom in 2003 than at the top in 1999， though the “returns-based” model indicated the opposite。One small addendum on Intel -- even with their own example， the authors could barely make their returns-based model work。 From 1991 to 1998， Intel grew revenue at a 28% annualized rate。 But assuming such a high organic growth rate would break the model。 So the authors instead arbitrarily assume that Intel’s organic growth rate from 1999 onward should be 8。5%， which is based on a 1999 GDP forecast of 5-6% and an assumed 3% “natural” organic growth rate。 This 8。5% growth rate literally had no relationship with Intel’s actual business performance at the time。 It was just shoe-horned in， because the authors’ model cannot handle anything but single-digit longer-term growth rates！I don't know how to put this any more clearly: the authors' "expected returns" model for evaluating growth stocks is useless for 90% of growth stocks， is useless even for the growth stocks that fit within the model's parameters， and leads to both nonsensical assumptions and nonsensical results in the author's own examples。As a final aside， I was also disappointed by the sheer number of errors throughout the book。 There are a litany of basic spelling and typographical errors。 There are errors in the mathematical formulas in the appendices。 There are errors related to the calendar dates quoted in the Intel example。 And then there are other inexplicable inconsistencies that further evidence the fact that the book has five authors working separately。 For example， in the WD-40 example， the author uses a “real”， rather than “nominal”， organic growth rate estimate。 In no other section of the book are “real” growth rates used。 Instead， the other examples always use nominal rates inclusive of inflation。 No explanation is given for this inconsistency。Ultimately， I think the first edition of this book was quite good。 I think this edition is significantly worse。 。。。more