Thursday, 5 November 2015

Whole Foods Muppet Math and the Perversion of Margin of Safety

Yesterday's post on Whole Foods got me thinking about a number of issues, including, but not limited to the dangers of using excel to justify a current valuation, instead of being used to stress test valuations. It seems that yesterday's article from 2005 was heavy on the former, and entirely absent on the latter. Coupled with convincing rhetoric, a fancy website, a subscription service and a best selling book, it's hardly surprising that investors as a whole make material errors in judgement (I'm no exception) listening to supposed experts (of which I am truly NOT).

I want to examine a number of the assumptions presented in the 2005 article and cut them to shreds. Not because I'm smarter or have a better track record than the author, I'm not, and I most certainly don't.  I want to highlight these assumptions in order to demonstrate the variability on output of changing the inputs to a valuation model.  Garbage in = garbage out (or so someone said).

I'm nowhere near an expert at any of this.  I try and approach ideas with at least four facets which I hope, will prevent me from doing something idiotic. These facets are:

  1. A healthy dose of skepticism
  2. Humility
  3. Common sense (I hope)
  4. Discipline
The most difficult of the four facets are #'s 3 & 4, why? Because faced with having to make a decision, my mind plays tricks on me.  Anytime a new idea seems to present itself, my impulsive reaction is to do something immediately.  This is where modelling comes into play.  The process of running an idea through a model seems to set me on a wait and see path as opposed to a buy first and model later path (although I've had plenty of the latter types of experiences along the way).

Let's look at Mr. Town's assumptions juxtaposed to stress testing his assumptions by 1/2:

Phil Town assumptionsStress test assumptions
EPS yr 02.32EPS yr 02.32
G = 22%G = Phil G / 211%
Future P/E yr 10442 x EPS GFuture P/E yr 10 = Phil P/E / 222
FV of EPS yr 1016.95FV of EPS yr 106.59
(2.32 x <1.22>^10)(2.32 x <1.11>^10)
Value per share yr 10745.66Value per share yr 10144.92
(16.95 x 44)(6.59 x 22)
Sticker price184.31Sticker price35.82
(745.66 / <1.15>^10)(144.92 / <1.15>^10)
MOS price = Sticker price / 292.16MOS price = Sticker price / 217.91
Current price90Current price90

A few things to note:

  1. The author establishes the tone at the beginning of the article by noting he wants not only a margin of safety, which is equivalent to his buying $1's for $.50, but as an additional cushion, he'll only accept a 15% rate of return.
  2. Given the above, why does the author accept consensus estimates regarding future growth as given?  Surely, if he were to err on the side of conservatism, he'd question future growth over the next decade.
  3. Given the above, why does the author use a P/E of 44 to extrapolate the past decade's growth (which included growth in store units from a very small base)?
  4. I agree with using a 15% required rate of return, but he's jumbled aggressive nonsensical inputs into a quick and dirty attempt at a valuation without asking the important "what ifs".  These what ifs are: a) what if G does not equal 22% "compound", and b) what if PE compresses from 60x ttm eps below 44x.
 You can see how he's used aggressive nonsensical inputs regarding growth and future PE to justify the current valuation.

Next to his valuation, I juxtaposed a simple analytic chopping G in 1/2, from 22% to 11%, and chopping P/E in 1/2, from 44 to 22.  Completely different picture isn't it?  In this case, I don't want to own Whole Foods in 2005.  In fact, I reckon at this point in its history, it's so severely overvalued, I want to short it.

I've done the same analysis today starting with current expected eps of $1.63 below.  If I want a 15% required return, what should I pay today assuming 5% growth in EPS, and a P/E of 15x, and a 50% margin of safety?:

Stress test assumptions
EPS yr 01.63
G = 5%
Future P/E yr 10 = Phil P/E / 215
FV of EPS yr 102.66
(1.63 x <1.05>^10)
Value per share yr 1039.83
(2.66 x 10)
Sticker price12.82
(26.55 / <1.15>^10)
MOS price = Sticker price / 26.41
Current price30

Another comment, if I want a margin of safety, wouldn't using a 15% Re already reflect this?  I would think so.

As I said before, garbage in, garbage out.  Ignoring the stupidity of the 1/2 x sticker price adjustment, what level of growth is the market currently pricing in today, assuming the future P/E is equivalent to 2G?

Stress test assumptions
EPS yr 01.63
G = 12%
Future P/E yr 10 = Phil P/E / 2242 x EPS G
FV of EPS yr 105.06
(1.63 x <1.12>^10)
Value per share yr 10121.50
(5.06 x 10)
Sticker price30.03
(121.5 / <1.15>^10)

Using common sense, we likely discount this scenario because we don't expect G @ 12% given competitive environment and margin pressures.  We also see that the P/E is nonsensical, as it's higher in 10 years than current (unless of course, earnings expand justifying a higher P/E).  Therefore, I expect that both G must be lower and the relationship between G and P/E is not exactly 2:1, and my required return should likely be lower than 15%.  It's a fairly stable business, we're not valuing VC cash flows here.  If I drop Re to 12%, drop G to 6%, and drop P/E to 15x:

Stress test assumptions
EPS yr 01.63
G = 6%
Future P/E yr 10 = Phil P/E / 215
FV of EPS yr 102.92
(1.63 x <1.06>^10)
Value per share yr 1043.79
(2.92 x 15)
Sticker price14.10
(43.79 / <1.12>^10)

About 50% below current.

Concluding Thoughts

All of the above is purely hypothetical,  You can drive yourself crazy trying to back into the numbers in order to justify a valuation (which probably explains a lot of the volatility and reactionary trading we see on a daily basis).

I happen to prefer using free cash flows to value known or likely cash flows.  At the end of it all, this is more art than science.



  1. Interesting analysis and you bring up a good point about testing assumptions across a range of future scenarios. I tend to forget that step from time to time but it's an important step to see the range of possibilities. After all the future is unknown so at best we're guessing, some better than others.

  2. I'm no stranger to shortcutting the process of analysis, myself! Hopefully something I can address and change over time with positive results...