Tuesday, 15 September 2015

American Express and Further Thoughts on Margin of Safety


Before I get going, a quick aside on what I'm attempting to accomplish here: I'm looking to empirically examine how much of a theoretical margin of safety certain Dow component stocks provide right here, right now as at September 15, 2015, with a view to constructing a portfolio over time.

In my opinion, the key consideration in constructing a portfolio is not how much yield a portfolio generates.  Yield is just about the last thing on my mind in evaluating investment candidates.  Rather, it is how much of a margin of safety does a portfolio provide each time I add incremental units (i.e., shares of different companies).  Not only do I want to achieve theoretical diversification of risk: this alone is not enough. Instead, I want to ensure that I am buying $1's for less than $1 ($.50 where possible) each time I add incremental units.  And until I find $.50 $1's, I should be content to sit in cash and do nothing.

Case in point, I could run a simple screen of lowest P/E, lowest P/B, lowest P/S, and highest dividend yield, and throw 50 darts, and I would end up with a nice portfolio of everything perceived to be yield oriented dividend champions, and which every other dividend growth investor in the universe also owns (albeit in different proportions). 

Well, you know what they say, misery loves company.  I could throw 50 darts today along these lines right now, end up with an average expected yield of 3.5% across the entire dividend champions universe in equal weight proportions, and guess what, I'd likely be no better off than if owned SPY.  So rather than incur 50 commissions on different 50 purchases today, why wouldn't I just buy equal weight SPY at regular monthly intervals over the next year?  At best, I'd save myself 38 lots of commissions between now and the end of one year.  At worst, whatever I give up in terms of yield, I make up in terms of being diversified.  My only remaining risk is systematic risk, which is non-diversifiable.  Empirically, with 50 different dividend champions, my expected correlation of risk and return is basically going to match SPY over time, so if the expected risk is the same, then my remaining risk is the risk of adverse price movement in either case.  Put another way, if SPY falls 50% over a two year period, my starting yield doubles.  Great.  My draw down is now only 43%.

In order to identify $1's for $.50, while at the same time avoiding cigar butts, I must concentrate my focus on the Balance Sheet.  I must also screen the 52 week lows daily (sorry folks, you are not going to find $.50 $1's on the daily new 52 week high list, I promise you this).

I'm going to borrow a concept from reading about Paul Sonkin's approach to evaluating 52 week lows, and the big secret is, he evaluates after tax EBIT / EV = cap rate.  From my readings, Sonkin looks for high after tax cap rates as a starting point for his analysis.  So what constitutes high enough?  I'm not sure yet.  I've been running a 52 week low screen for almost a month now, and I'm seeing lots of cap rates in excess of 15%.  I've seen as high as 40%, but these seem to be businesses which are severely lacking in terms of earnings sustainability (case in point, Torstar came up at 70%).

Sonkin also does not concentrate his focus on large caps, so hopefully I'm not trying to fit a square peg into a round hole by borrowing his methodology and applying it to large, mid, small, and microcaps.

With this in mind, I'm also going to ignore certain Dow component stocks which do not, in my opinion, deserve consideration, due to lack of sustainable franchise/moat.  For the record, I am going to ignore:

  • Caterpillar
  • Chevron
  • Dupont
  • Exxon

I'm not going to waste my time evaluating companies which do not have pricing power, and/or do not have a definable moat.

So far, I've looked at Procter and Gamble and Johnson and Johnson (I haven't posted by JNJ analysis).

Now onto American Express.

First off, my EPV analysis.  For this, I've made some simplifying assumptions.  First, unlike Visa or Mastercard, Amex assumes credit risk.  The balance sheet is levered up as a result.  From my readings, they make money in one of two simple ways.  Charge card discount and other fees, and member loan interest (spread).  They also make money from FX on charge business, travel commissions, and other incidental business, but by far, the largest revenue streams are discount, loan interest, and member fees.

When I analyzed the Balance Sheet for the purpose of determining EPV, I made simplifying assumptions that the card-member A/R should be netted against short-term deposits (at Amex bank), and that the net loans should be netted against the Long Term Debt. under the presumption (rightly or wrongly) that the overall loans/debt are a function of the business of issuing cards and granting member loans.  Were they not in this business in the first place, they wouldn't have borrowed $100B!

I made these assumptions for a simple reason: when I first calculate EPV (which is adjusted after tax EBT plus/minus net maintenance capx), and I then capitalize EPV under a range of WACCs, I have to get from firm-EPV to equity stakeholder EPV.  If I simply deduct 100% of the debt, I get a negative EPV, which doesn't make sense.

Here is my EPV analysis:

12/31/200512/31/200612/31/200712/31/200812/31/200912/31/201012/31/201112/31/201212/31/201312/31/201412/31/2015
Sales20,58922,81323,39022,56719,21025,61228,85029,59230,86432,24831,7134.41%
EBT4,2485,3285,5563,5812,8415,9646,9566,4517,8888,9919,0177.82%
EBT %20.63%23.36%23.75%15.87%14.79%23.29%24.11%21.80%25.56%27.88%28.43%22.68%
Avg EBT4,6695,1745,3055,1184,3575,8086,5436,7117,0007,3137,192
Add special charges00000000000
EBT before special charges4,2485,3285,5563,5812,8415,9646,9566,4517,8888,9917,1925.41%
Special charges as % of sales0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%
Deduct avg special charges00000000000
EBT after avg special charges4,2485,3285,5563,5812,8415,9646,9566,4517,8888,9917,1925.41%
SG&A13,44311,58113,25513,45111,54714,21016,18216,56816,45817,16816,558
SG&A as % of sales65.29%50.76%56.67%59.60%60.11%55.48%56.09%55.99%53.32%53.24%52.21%56.25%
R&D00000000000
R&D as % of sales0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%0.00%
Add back 25% of SG&A3,3612,8953,3143,3632,8873,5534,0464,1424,1154,2924,140
Add back 25% of R&D00000000000
Adjusted EBT7,6098,2238,8706,9445,7289,51711,00210,59312,00313,28311,3324.06%
Tax rate34.00%34.00%34.00%34.00%34.00%34.00%34.00%34.00%34.00%34.00%34.00%
EBT AT5,0225,4275,8544,5833,7806,2817,2616,9917,9228,7677,479
+ Dep'n6026456487121,0709179189911,0201,0121,011
-FCInv-608-766-883-977-772-887-1,189-1,053-1,006-1,195-1,442
Adj EBT AT5,0165,3065,6194,3184,0786,3116,9906,9297,9368,5847,0483.46%
NI per F/S3,7343,7074,0122,6992,1304,0574,9354,4825,3595,8855,9924.84%
$ Diff1,2821,5991,6071,6191,9482,2542,0552,4472,5772,6991,056
% Diff25.55%30.14%28.60%37.49%47.77%35.71%29.40%35.32%32.47%31.44%14.98%
Div's0.480.570.630.720.720.720.720.800.661.241.3310.73%
Cost of capital rates:EPVAdj for Debt & Excess CashAdj EPVO/S sharesEPV / shareBVPS / shareExcess EPVCurrent pricePrice / EPVPrem/Discount vs EPV
Upper bound - VC15%46,985.801658563,570.80101362.7521.842.8776.331.2221.63%
Arbitrary 2 <10% - 12 %>10.00%70,478.711658587,063.71101385.9521.843.9476.330.89-11.19%
Arbitrary 17.50%93,971.6116585110,556.611013109.1421.845.0076.330.70-30.06%
Per GF6.45%109,269.3116585125,854.311013124.2421.845.6976.330.61-38.56%
Combined NI & Div growth6.03%116,879.6716585133,464.671013131.7521.846.0376.330.58-42.07%
Adj BetaRfRisk Premium%D%EAT Int cost
1.032.19%4.81%20%80%1.90%


As you can see, at a range of WACC's between 10% and 7.5%, I have a theoretical margin of safety of between 11% and 38%.  Not bad to start, but here's the caveat: The EPV analysis assumes that starting EBT is constant.  More on this later.


Here's my Balance Sheet Reproduction Analysis.  Again, I've made a simplifying assumption by netting A/R and loans (rightly or wrongly).


AssetsAs reportedAdjustmentReproduction cost
Cash21,071100%21,071
A/R9895%93
Loans16,09585%13,681
Deferred tax assets00%0
Prepaid expenses0100%0
Other current 00%0
Total current37,26434,845
PPE4,01650%2,008
Intangibles0100%0
3 yrs SG&A and R&D49,674100%49,674
Equity method investments0100%0
Other investments4,582100%4,582
Other assets11,382100%11,382
Trademarks0100%0
G/W0100%0
Total LT69,65467,646
Total assets106,918102,491
Less:
Non-interest bearing liab's-11,056
Excess cash-20,754
Equals total net reproduction cost70,681
O/S shares1,013Greenwald Three States
Reproduction cost/sh70State One:
EPV/sh (avg)106AV > EPVAV = EPVAV < EPV
EPV less Reproduction cost/sh37No franchiseEquilibrium Franchise
Liab's
Bank advances54,877100%54,877
ST debt0100%0
A/P and accruals11,056100%11,056
Tax payable0100%0
Provisions0100%0
Other liab's16,594100%16,594
Defd rev0100%0
Total current82,52782,527


Next, my P/E Payback Analysis:

So far so good, Amex looks like it needs to only grow EPS by 4% cumulatively over the next 10 years in order for me to get repaid my initial purchase of $76.40 in terms of earnings.  Again, this analysis assumes that starting EPS stays constant and grows from $5.69.

TickerPriceEPS yr 2015P/EP/E yrs
AXP76.45.6913.4310 year payback
g?10
4%AnnualCumulative
20155.695.69
20165.9211.61
20176.1517.76
20186.4024.16
20196.6630.82
20206.9237.74
20217.2044.94
20227.4952.43
20237.7960.22
20248.1068.31
20258.4276.74
20268.7685.50
85.50


Finally, my comparative analysis across peers:

You can see that Amex, Discover and Capital One (credit card issuers who assume credit risk) trade at (and have always traded at) noticeable discounts from Visa and Mastercard.  The market values Visa and Mastercard at almost 2x EPS and almost 5x revenue, vs. Amex, Discover, and Capital One.  This has to be a function of participant preference for non-assumption of credit risk.  Funnily, the longer this goes on, and the more preference there gets for Visa or Mastercard relative to Amex, the riskier Visa and Mastercard get in terms of relative valuation.  The preferential valuation gets stretched to the point of idiocy, and a reversion occurs (at some point).  It's an oxymoron, the "safer" business model results in a riskier stock.  Who would have guessed?

CompanyAXPVMADFSCOF
Price76.3970.5292.3552.7776.65
Market Cap75,493170,763103,66022,84541,283
EV114,400155,525102,00538,15279,384
Net (cash) / debt38,907-15,238-1,65515,30738,101
Revenue (ttm)33,67013,5389,5398,57322,771
P/S2.2412.6110.872.661.81
EPS (ttm)5.692.533.264.847.07
P/E (ttm)13.4327.8728.3310.9010.84
PE (09 low)5.2627.0023.002.606.81
P/E prem/disc to 09 low255.23%103.24%123.17%419.34%159.20%
EBTDA (ttm)10,0288,8205,4133,8887,977
EBT (ttm)9,0178,3335,0403,5075,980
EV / EBITDA (ttm)11.4117.6318.849.819.95
EV / EBITDA (09 low)6.8015.0012.85-2.008.44
EV / EBITDA prem/disc to 09 low167.77%117.55%146.65%-490.64%117.91%
EV / EBIT12.6918.6620.2410.8813.27
EV / Sales3.4011.4910.694.453.49
Yield % 1.42%0.65%0.68%1.98%1.85%
EBT Margin26.78%61.55%52.84%40.91%26.26%
EBTDA Margin29.78%65.15%56.75%45.35%35.03%
10 year rev g4.60%0.00%12.50%-17.00%0.40%
10 year earnings g7.20%0.00%0.00%-15%4.60%
5 year rev g9.30%25.00%16.00%8.20%4.60%
5 year earnings g24.00%28%23.00%25.30%34.80%
Current ratio1.041.741.591.611.02
Int coverage5.47#DIV/0!82.622.923.79


So far, the results of my initial analysis are pointing me towards Amex, but wait there's more:

From the 10K:

  • 112.2m cards in force
  • $1T in spend globally
  • Largest source of revenue is discount fees.  In 2014, discount fees were $19.493B.  Working backwards, net discount fees are $19.5B / $1T  = 1.95%.  These are net of sharing with bank issuers of Amex cards.
  • Second largest source of revenue is interest on loans.  In 2014, loan interest was $6.9B.  Working backwards, effective interest rate = $6.9B / $69B = 10%
They generate about 30% of charge volume through co-branded business (i.e., Costco, Delta, Starwood, etc).  Costco was by far, the largest of the co-branded partners.  Costco and Amex terminated the agreement effective March 31, 2016. Amex’s explanation was that the upfront cost of the co-branded relationship was too much, and would not have resulted in long term economic returns for Amex shareholders.  Apparently, they compete with other issuers (I’m guessing Visa, Mastercard, etc) to establish co-branding partnerships with large companies such as Costco which could represent a significant volume of repeat charge business.  As competition for co-branding increases, the cost of signing exclusive agreements gets bid up.  Amex would have had to have paid Costco $X up front to continue the agreement (I’d like to now read Costco’s 10K to see what their side of the story is). 

  • Costco generated 8% of global charge business in 2014
  • Costco member loans accounted for 20% of total member loans at end of 2014
  • Amex sold its travel business into a Joint Venture in 2014, so the travel commissions line will disappear from the income statement going fwd


Recasting the P/L going forward to account for these changes:

  • Costco volume @ 8% x $1T  x 1.95% discount fees = $1.6B
  • Costco loans @ 20% x $69B x 10% = $1.38B
  • Travel commissions disappear = $2.5B (*** see comment below)


Take 2014 total revenue of $34.292 – the above $5.480 = $28.8B

Net margin in 2014 was 17%, so applying to the above, I get 28.8B x 17% = $4.9B, or $4.70 per share.  

At $75 per share, this is a fwd P/E of 16x, a far cry from the 13.3x expected forward earnings per share per gurufocus or Yahoo finance.  Yahoo finance has the lowest estimate at 5.21 next year, and thinking about this logically, my forecast is pretty conservative.  However, the travel commissions aren’t going to disappear entirely, Amex is going to share them via a joint venture using the equity method pickup (non-consolidated), so it’s probably incorrect of me to eliminate these entirely.

Even so, you can appreciate how significant the Costco business was in terms of P/L impact above.  Now, maybe, signing up Sam’s Club will mitigate some of the loss?  I’m not sure.  Sam’s club has often been compared as second tier to Costco, but it’s still a significant player.

Let’s say I’m right and EPS at the tail end of 2016 after the demise of Costco is forecast by management at $4.70, not $5.21.  This is how we potentially get to $60 from $75 currently!  I have a model target price of $56.70.  This is my buy price.

Here are the challenges as outlined in the 10K

  • Strong USD putting pressure on global revenues (translated back into USD)
  • Department of Justice point of sale discrimination appeal (could limit their ability to force merchants to honour Amex exclusively)
  • New regulations limiting merchant card fees, will drive down discount revenue
  • Competition for charge business will drive down fees
  • Competition for co-branded business will drive up cost of co-branded business
Here are the strengths as outlined in the 10K
  • Continued shift away from cash to charge, prepaid, or similar
  • They own the closed merchant network (possible non-recognized asset), it has value, and they can use it to analyze spending habits
  • Undeserved niches in financial services
  • Goodwill and reputation
  • Big Data expertise in risk mgmt., marketing, and servicing

Certainly, the cheaper Amex gets, the more compelling an idea it becomes, and at $56.70, anyone who ponied up at $92.5 in mid-2014 is long gone.

Finally the long term monthly chart:



No comments:

Post a Comment