Saturday, 26 December 2015

Revisiting Model Assumptions Using Procter & Gamble (Again)

When I first started blogging this summer, I was pretty gung-ho about having discovered Bruce Greenwald's writings.  I quickly built my own EPV valuation model based on his teachings, without thinking too carefully about the model inputs or outputs.  Well, Greenwald would probably be the first person to say, garbage-in, garbage out, and my model valuations were no exception.  As I enter 2016, I have more questions about the models I've strung together than answers (which is a good thing, I think).

I chose PG to analyze back in the summer because the company is a useful barometer of blue-"chipness" in general.  It's a dividend aristocrat (whatever that means), a key Dow component, and is most likely a key component in any serious dividend investor's portfolios.

Here's a snapshot of PG's current stats courtesty of gurufocus:

I've highlighted a few areas of concern right off the bat, on the summary above, which require further investigation / clarification:

  1. The reported ttm P/E is 31x.  This seems high for a steady, slow growth consumer products company.  The curious investor needs to ask him/herself why the ttm P/E is 31x.
  2. Similarly, the NRI P/E (non-recurring items) is pretty high as well, at 26x.  Same comments apply, why so high?
  3. Similarly, P/S seems high at 3.1x.  Same comments apply, why so high?
  4. Similarly, price to FCF as reported seems high at 20.77x (on the subject of P/FCF, another criticism, why use P/FCF?  Shouldn't we use EV / FCF? After all, price is just market cap, EV is enterprise value, which belongs to both bondholders and  shareholders.  The fact that P/anything is so widely used and accepted off the cuff is actually somewhat of a joke)
  5. Finally, EV / EBIT seems high at 19x.  Same comments apply, why so high.

The recurring theme in 1-5 above may not be all that straightforward:  Either PG is currently overvalued causing all cursory (and widely accepted) valuation ratios to seem high, or all measures of value (EBIT, earnings, sales, FCF) as published (scraped) are wrong, or if not outright wrong, non-representative of actual earnings power on a go forward basis.  While there may be some truth to the numbers, the job of the curious investor is to uncover the truth.  

But wait a second, gurufocus, yahoo finance, google finance, marketwatch, morningstar, etc., all show the same (or similar) valuation ratios and the same (or similar) published numbers.  If all of these unbiased sources of information have the same or similar published numbers, how can they be wrong?  Well, the longer I actively research, the longer I realize that it's my job to ignore published numbers from all sources except for the company itself.  Why?  Because an algorithmic scrape is a just a scrape, and it doesn't read the 10K or 10Q or MD&A for explanations regarding the published numbers.

Case in point, here's PG's summary and brief discussion of 2015 reported results:

I've highlighted some key areas of focus above:

Net sales were indeed down by $4.1B in 2015 vs. 2014, in USD terms.  This was due to foreign exchange.  Whether the weakness in all other currencies vs. the USD persists in the short term vs. medium term vs. long term is anyone's guess, but I ask this: if PG continues operating as is for the forseeable future, with 37% of revenue derived from the United States, and 63% of revenue derived from outside the United States, and PG does not repatriate its international cash or dispose of its international operations and repatriate the proceeds, has PG really experienced loss?  Yes and no.  On paper, yes, the value of international sales has dropped relative to USD's, so on FX translation of foreign subsidiary sales, GAAP requires that sales get translated at avg FX vs. USD for the period, which in the case of 63% of International sales,  dropped vs. the USD.  No, for the simple reason that I don't believe PG as currently structured is about to readily dispose of a significant portion of its International Operations. Here's PG's statement of OCI for the year.  You can see how large the FX translation adjustment was for 2015 relative to net income:

FX translation losses were $7.2B, almost 100% of reported net income!

We can make an eduated guess that the FX translation loss getting added to OCI and included as part of equity is a result of foreign self-sustaining subsidiaries reporting in their own functional currency.

So, in the above case, the constant un-translated dollar amount of revenue would likely be about $4.1B higher if not for the extreme FX moves which occurred during the twelve months ended June 30, 2015 (and which have continued thereafter).

The next key area of focus is also FX related.  Per note 1 on the slide above, the company explains that net earnings were negatively impacted by $1.4B due to FX, $2.1B due to non-cash impairment charges incurred on disposal of the Duracell batteries business, and a $2.1B charge due to deconsolidation of the company's Venezuelan operations.  In total, about $5.6B of charges not expected to recur, and not typically tied to the ongoing business of selling diapers, or laundry detergent (not accounting for restructuring charges).

Recasting the ratios above:

Sales of $76B + $4B FX = $80B
Net earnings of $7B + $5.6B (assuming after tax) = $12.6B 
EBIT of $11.8 + $2B + $1.4B /( 1-.25) = $15.67B

Adj P/E = $217B/$12.6B = 17.2x ttm
Adj P/S = 2.71x
Adj EV/EBIT = 15.1x

Not nearly as bad as originally thought.  Certainly, 17.2x ttm earnings is a lot more palatable than 31x ttm earnings!

What about overall valuation?

For this piece of the exercise, I had to revisit my model.  As I was grabbing the reported numbers from the last 10 years of annual reports, I ran into a problem.  Namely, the reported results don't stay constant! Don't believe me?  Compare the 10 year summary of selected financial info from the 2015 annual report to the 10 year summary of selected financial from the 2013 annual report, reproduced here:

The difference in reported results is due to retroactive restatement of historical results in 2015 attributable to disposition of operating segments which were historically consolidated in previous years (no one said this was easy).

So how does one analyze a company  in which the reported revenue base is constantly changing? The answer, very carefully!

The more I think about this, the more I think that PG is in two businesses.  One: consumer products, and Two: acquisition and sale of consumer product businesses.

The second piece is evident from a review of the cash flows from investing activities each year, as so:

I'd hazard a layperson's guess that purchase and sale of consumer products businesses is likely going to continue as a key part of PG's ongoing strategy going forward, so I don't think I can ignore the net cash flow impact in evaluating the overall business, however, I don't want to understate the impact of PG having to invest in additional capx (more on this later).

Here's the company's P/L as stated, and adjusted by me for the last 10 years:

A few observations:

  • Gross margins have remained at or close to 50% over the last decade, indicative of sustainable pricing power.  This isn't surprising given the defensive nature of PG's products
  • EBIT margins have remained at or close to 20% over the last decade, but have trailed off over the last few years.  On further investigation though, the reason for the diminished EBIT margins recently seems to be a function of PG's restructuring program, initiated in 2012.  A cursory review of diminishing EBIT margins without asking why could potentially lead to incorrect conclusions regarding the sustainability of EBIT margins.
  • Between 2012 and 2015, the company has incurred both restructuring charges and taken non-cash charges of close to $9B.  As a mature company, my guess is that in order to drive growth in the face of intense competition for consumer budgets, the company has had to resort to internal cost savings programs.
  • SG&A as a % of sales has remained fairly constant over the last decade.  This isn't surprising given that PG is a marketing machine.
  • For now, I haven't arbitrarily added back a % of SG&A / Advertising expense in attempting to determine EPV.  I will come back to this shortly though

Here's the remainder of the recast P/L:

My biggest issue with the recast P/L is what to do with capx.  If I include capx - cash proceeds on sales + acquisition of businesses, I can get into a situaton whereby I overstate current year's normalized after tax EBIT.  If I ignore - cash proceeds on sales + acquisition of businesses, I may end up understating the current year's normalized after tax EBIT.  So, I took the average net capx over the last 10 years.  

I've also included an as-stated Dupont Analysis of ROE in my model, but I don't think it's useful as is, because it uses reported NPM as one of its inputs.  What I should be doing in order to properly analyze ROE is recast NPM using the after tax-impact of non-recurring charges I've identified. Here's my recast Dupont Analysis:

As can be seen, out of all of the component parts comprising ROE, Asset Turnover and Financial Leverage have been fairly constant over the last decade (with the exception of 2005/06 during which PG added Gillette, signifincantly increasing the asset and equity base as a result).  Interestingly, recast NPM has remained fairly stable between 14.5% and 15.5% since 2011.  This leads me to believe that the company is targeting ROE of around 18%.

Here's my first crack at valuation using my EPV model:

And, not surprisingly, my first comment regarding the valuation is that it's wrong, because it doesn't take brand value into account.  Unlike my previous attempt at valuing AXP, there is no published brand value for PG as a whole using Brand Finance.  However, Brand Finance does list values by individual brand, see attached:

Per above, Gillette and Pampers alone have estimated brand values approaching $15B. Unfortunately, Brand Finance doesn't list every single brand offered by PG, but I don't think I really need to know what every single brand is supposedly worth in order to come up with a reasonable valuation estimate.  I just need to know that a value exists, and it's likely greater than the $15B attributable to just these two brands.

I further thought that a reasonable proxy for fair value of brand worth might be reported goodwill. After all, it represents the excess paid for fair value of identifiable net assets on acquisition by PG, and as PG is fairly acquisitive, Goodwill might be a relevant benchmark to use for brand value.

Here's a quick table showing Goodwill as reported, and as adjusted by me:

Goodwill carried:44,61233,45916,730
o/s shares2,8682,8682,868
g/w / share15.5611.675.83

I'm going to rule out a 50% reduction in reported Goodwill on the basis that just two of the brands per Brand Finance were estimated to have brand value of $15B, which would suggest that across close to 40 brands, PG's remaining brand across the remaining brands is about $1.7B (which is absolute rubbish).

If I take 100% of goodwill as indicative of total brand value, I have to estimate the correct %ge of SG&A & R&D to capitalize in order to bump EPV / share by around $16.

Here are the results:

I determined that brand value of $44B is equivalent to capitalizing approximately 30% of SG&A and 25% of R&D, which is consistent with Greenwald's teachings.

Final Thoughts

I'm hoping that I've demonstrated how important it is to not take reported numbers as per mainstream financial reporting sources as given.  I'm also hoping that I've demonstrated how thinking critically and recasting operations is essential in developing an investment thesis.  Caveat, none of the above suggests that PG is currently cheap.  If actual WACC is somewhere between 8% and 10%, then PG is at best, fairly valued, and at worst, about 20% overvalued.

Interestingly enough, here's where PG got during the July/August 2015 lows:

And here's the valuation table at $65...all I can say is that it's not surprising that PG found support here...


  1. Nice workout there, I really enjoyed the inquisitive approach.