Thursday, 24 December 2015

2015 Dogs of the Dow: Revisiting American Express, Paccar, and More on Mike Burry

More on Mike Burry.  Last post, I attached a link to Burry's writings from 2000/2001.  Here's the link again, and an excerpt on his analysis of Paccar, circa 2000.

(I'll attempt to tie the concepts back to American Express shortly)

I think the key takeaways here are as follows:

  • Don't take reported mainstream financial ratios/metrics at face value, especially when screening for ideas.  In Paccar's case at the time, Yahoo Finance reported a debt:equity ratio which was misleading in the context of understanding the entire business.  Paccar's overall business was (and still is) two seperate businesses rolled into one.  A trucking and parts business, and a finance business.  In Paccar's case, the debt was (and still is) allocable to the finance business.  The trucking and parts business actually didn't (and still doesn't) have any allocable debt at the time.  Currently, Paccar has about $3.5B in cash and securities allocable to the trucking and parts business, but minimal debt
  • This mode of analytical thinking can certainly be extended to examining a wide range of businesses today.  At first thought, I think of companies with integrated finance businesses. GE comes to mind first and foremost, as does Caterpillar.
  • The investor's job is to stratify the distinct businesses if possible.  This is not an easy task, but I think it's necessary in order to do a proper valuation.  If the sum of the parts, conservatively valued, exceed the whole (being current enterprise value), there may be a margin of safety.
  • I find it fascinating that Burry concluded that he was ignoring the finance company debt in evaluating Paccar at the time, but it sort of makes sense.  The debt was tied to and supported the loan book (ignoring the debt worked up until, oh, about 2007/2008).  To be honest, I don't think an investor can ever fully ignore debt, but because the loan book showed an equity/assets ratio of 20% vs. comparable commerical bank ratios of between 5% - 8%, Burry concluded that the loan book was relatively safe
  • The ratio of Equity / Assets is extremely important in analyzing financial entities (or financial operations within an entity)

What does any of this have to do with American Express?

American Express comes to mind for a number of reasons.
  • AXP (the stock) has performed terribly in 2015 relative to the broader market.  On a YTD price performance basis, it's right up there with Walmart and Caterpillar.  See pic below:

  • The company is facing numerous challenges:  Merchant fees are getting squeezed, Costco severed ties this year (impact yet to be felt), the EU and DOJ could potentially impose restrictions on merchant exclusivity, Visa and Mastercard are taking market share in terms of charge business, customer charge habits are constantly changing, foreign exchange headwinds are adversely impacting international charge business, etc. (the more bad news I read, the more bullish I get).  Here's the outlook provided by the company for the remainder of 2015 through 2017:

  • In response to these challenges the company has attempted to ramp up spending on marketing and has engaged in restructuring efforts.  The company has also attempted to sign additional co-branding agreements, most recently, Sam's Club.
  • The company certainly seems to fit into the analytical company within a company Burry evaluation framework.  Here are the most recent statistics, courtesy of gurufocus:

Notice the huge discrepancy between market cap and enterprise value, about $30B, largely attributable to the company's debt in excess of cash and securities.  But hold on a second, what if we attempted to stratify the charge business from the loan/credit business?  (Incidentally, this exercise is purely hypothetical on my part).

Charge Business

From my understanding, the charge business is driven mostly by merchant fees and services, customer credit card fees, and travel related services. 

For this piece of the analysis, I took the last 10 years of reported "non-interest" revenues and stratified the P/L as follows. I had to make number of assumptions here in terms of matching up the appropriate non-interest related categories in the P/L.  Here are the results:

A few observations:

  • This is a low growth business in terms of revenues.  Charge revenues have really only grown 2.8% over the last decade.  I don't realistically expect this to change, although revenues could likely grow at a slower pace going forward given the challenges noted above
  • Notice the drop-off in charge revenues between 2008 and 2009.  Charge business, travel business, etc., are highly cylical and sensitive to economic disruptions.
  • Notice that EBIT as a % of revenues has been steadily dropping since a decade ago.  This could be symptomatic of diminishing discount rates.  I've taken the average EBIT %ge of 16.18% to recast the P/L over the 10 years studied
  • SG&A as a %ge of revenue is significant.  This isn't surprising given the company's recently implemented initiatives to drive brand recognition.  
  • For this initial analysis, I haven't added any %ge of SG&A spend back in order to hypothetically capitalize SG&A spend, although an argument can certainly be made that some portion of SG&A / marketing efforts should be capitalized in an attempt to determine sustainable earnings power value (more on this later)
Here's my initial attempt at determining a valuation range for the Charge Business

A few more observations:
  • This is only part of the picture.  I'm not putting much credence into the premium/discount to EPV calculation, as the valuation only reflects charge, and not loan
  • I've assumed that the majority of the debt is allocable to the loan business, and that the net cash excess cash on hand is a function of Basel capital tier ratios required 
  • The cost of capital rates are estimated ranges 

Loan Business

The second business which accounts for most of the carried debt, is the loan business.  For this business, I analyzed both the last decade of reported results allocable to the loan business, and I analyzed the ratio of equity to assets in each year.  Here are the results:

A few observations:
  • Loan is also a low growth business in terms of top line, just under 4% CAGR over the last decade
  • EBT as a %ge of Interest Income is significant at between 45% and 53%.  This may be more a function of my not allocating any amount of overhead or SG&A to Loan, but what I haven't allocated to Loan, I've allocated to Charge, so the end result will be a combined valuation of the two pieces which reflects the overall correct SG&A (again, this is purely hypothetical, and I haven't looked into each year's segmented results to see if the company breaks the actual amounts out)
  • 2007 through 2009 really took the company by surprise.  Observe the sheer scale of the loan losses in 2008!  Perhaps the company fell victim to similar delinquencies in its loan book as every other money centre bank experienced?  I can't recall if Amex had a mortgage business in 2008, but if it didn't, perhaps it held subprime paper and had to write it off?
  • If I call 2007-2009 an anomaly, maybe the current loan loss provisions are more precise?  I've reduced EBT down from current of $3.9B for the annualized ttm ending 12/31/15, to $3.38B, however, this doesn't mean that the loan loss provisions relative to the scale of the loan book are correct
  • The company doesn't give net income by segment, so I've estimated net income using the overall tax rate
  • The equity / asset ratio has improved signficantly from the depths of 2008/2009, but is still well below 2005 at 14.9% vs. almost 28%
  • The company's loan book has grown from $41B in 2005 to $69B currently.  Card member A/R has grown from $36B in 2005 to $46B currently.  Both pieces have grown commensurate with growth in debt and deposits.  
  • The larger the overall debt grows relative to homogeneity in the loan book, the riskier the company potentially becomes

Here's my attempt at valuation:

Tying it all together 

So I now have to seperate valuation tables for the two businesses, one for Charge and one for Loan.  I combined the two businesses into a consolidated P/L, and came up with the following combined valuation:

Problems with the combined valuation

Right off the bat, I noticed one problem.  American Express as a whole, has a brand value which hasn't been reflected anywhere above.  When I initially read Bruce Greenwald's book, he suggested capitalizing 25% of SG&A  / marketing in determining normalized earnings power value.  I haven't done this, because I don't know if 25% is right (it seems too arbitrary).

What I do know is this:  If I'm ultra conservative and do not capitalize any SG&A, I'm likely undervaluing the company.  And if I make the mistake of undervaluing the company, I may not determine its viability as a candidate for purchase correctly.

So here's how I attacked the problem:

I found a ranking of the top 500 most valuable brands, see link here at Brand Finance.  No one would be surprised to find Apple at the top of the ranking.  American Express was ranked 38th (down from 33rd in 2014).

Despite the drop, the company still seems to have brand value of around $21B according to Brand Finance.  Ok, who the hell is Brand Finance, and why should I listen to them?

If anyone's interested in reading about their ranking methodology, you can find it here.

Let's assume I should, and the brand is worth $21B, or about $21 per share.  This changes my valuation.

Assuming a 12% stress test cost of capital, I get $59.47 capitalized no growth combined EPV +$21 in brand value, for an estimated all in valuation of $80.53, about $10 higher than current.

What if I don't trust Brand Finance and I want to cut their estimates in 1/2?  I'd get $10.5B in brand value, or about $10.5 per share.  Again, this changes my valuation.

In this case, assuming a 12% stress test cost of capital, I get $59.47 capitalized no growth combined EPV +$10.5 in brand value, for an estimated all in valuation of $70, just equal to current.

If I believe that actual real cost of capital should be lower than 12%, perhaps between 8% and 10%, my valuation changes again.

At 8%, and $21B in brand value, I get an overall valuation of $100 per share

At 10%, and $21B in brand value, I get an overall valuation of $88 per share

And so on, and so on, and so on...

Concluding Thoughts

As usual, this exercise is more art than science.  The above excercise is not a buy recommendation, rather, it's a sit up and take notice suggestion.  There is nothing saying that American Express won't continue lower on the heels of continued worry in the market over continued loss of market share, DOJ / EU restrictions, and the "inevitable" demise of the legacy merchant fee business in favour of competitors.  But I suspect that the more bearish the story gets, and the lower the stock goes, the better an opportunity it becomes.

Quick Update December 25th, 2015 (Merry xmas)

After I finished this post last night, I asked myself what capitalized %ge of SG&A is equivalent to $21 per share in brand value?

I can use my model as constructed in order to detrermine changes in EPV based on changes in add-backs in SG&A, and then make a determination as to whether the add-back %ge seems reasonable. To me, this seems more intuitively appealing than just using a straight 25%.

Interestingly, in order to create $21 in EPV due to brand value, I'd have to capitalize 35% of SG&A.

Here are the results:

The takeaway here is that capitalizing SG&A / marketing spend of about 35% results in recognition of brand value of around $21 vs. my previous attempt at valuation where I capitalized no SG&A / marketing.

Do I think 35% is reasonable?  Maybe?  The truth is, I don't know.  I do know that the company spends a substantial amount on marketing every year ($10.9B currently, up from $5.8B a decade ago), and I don't believe I'm arbitrarily capitalizing spend that shouldn't be capitalized.  The company tracks marketing spend as a seperate line on the P/L called: "Marketing, promotion, rewards and Card Member services".  I've only considered this line in my analysis.

If I want to be conservative, and want to chop $21B in brand value in 1/2 (perhaps reflective of eventual dimminution of the brand in favour of VISA  or Mastercard, etc), here is the resulting valuation using $10.5B in brand value:

In this case, marketing spend is only about 18% effective in terms of capitalization, and my margin of safety across a range of different cost of capital rates falls vs. 35% capitalization.

This brings me to more concluding thoughts / questions, in no particular order (and I apologize for this in case anyone is looking for a recommendation):

  • I'm using approximations of WACC here.  To be fair, I don't think WACC should be as high as 12% for an established company such as American Express (although one could make an argument that during 2007/2008, the guys from the Big Short would have argued to the contrary and they would have been right)
  • Maybe I need to use different WACC's for the different businesses.  Charge is inherently less risky than Loan.  Maybe the answer is, model Charge at a much lower WACC, given that the business is predicated on Cards if Force, merchant services, discount fees, etc, all of which don't typically involve lending activities, and model Loan at a higher WACC to account for the potential in a spike in loan losses
  • I'm not sure whether 18% or 35% capitalized marketing spend is correct in terms of recognizing brand value.  One thing I am certain is that there is some value and it's significantly higher than 0%.  If it weren't, we'd be into an impairment model in evaluating the business, and I don't think this is the case.
  • I'm more preferential to the 18% capitalized marketing spend scenario because it's conservative, but in doing so, and then by using a high discount rate, am I double dipping when I shouldn't be? i.e., if I've already modelled zero growth, conservatively adjusted EBIT / EBT, and used 1/2 the value of brand value, should I also be looking at the stress test case?  My thoughts here, are that if price moves significantly below my stress test valuation discounted at 12%, either I'm getting a very good price, or there is something fundamentally wrong with the business (or everyone else in marketland has lost their collective minds by bidding Amazon up to the moon in favour of binning AXP)

No comments:

Post a Comment