Amazon.com
Introduction:
We
have long taken issue with consensus analysis of Amazon.com’s (AMZN) financial
statements and the consequent discounted cash flow (DCF) based valuations. Largely
due to management’s ability to successfully skew the discussion of free cash
flow ((FCF)) toward its preferred calculation, most DCF models ignore material
economic costs from the company’s stock-based compensation (SBC) program and
overstate the economic benefits of its retail business’s negative working capital
cycle. As a result, although we have long been both satisfied customers and
believers in AMZN’s ability to gain long-term market share, except for rare
occasions, we have concluded that the stock price over-valued the company’s potential
discretionary FCF.
During
the past seven quarters, an additional factor emerged that further questions
consensus equity valuation: As management has aggressively pursued its various
growth initiatives, capital intensity has exploded higher into increasingly
short-lived assets. Critically, management has chosen to finance much of this
equipment through capitalized lease commitments. However, as we will discuss in
detail below, the interplay between the company’s definition of FCF and
Generally Accepted Accounting Principles ((GAAP)), leaves the cost of these
leases unaccounted for.
As
long as AMZN’s operating performance remained in line with consensus
expectations, our insights into the company’s valuation presented no actionable
trading ideas. However, given the company’s disappointing third quarter
earnings report we have been surprised that our concerns have not found a wider
voice. Rather, after the company’s third quarter report, two prominent analysts,
Katy Huberty of Morgan Stanley (MS) and Mark Mahaney of RBC Capital Markets
(RBC), released aggressive purchase recommendations, with identical target
prices of $420. With the shares now trading above their pre-earnings price, we
believe our analysis presents a timely, alternative perspective.
We
have reviewed the reports referenced above. In each case, the target price
requires heroic assumptions for long-term revenue growth and margin expansion.
More importantly, we believe that each report also ignores the company’s exploding
capital intensity and misrepresents actual discretionary FCF accruing to
shareholders.
As
a rejoinder, we will attempt to reset the bar, as it were, by properly reflecting
the company’s existing cash flow dynamics and economic costs. To accomplish
this task, we present analyses of three factors; (1) The actual economic cost
of the company’s SBC plan, (2) a reasoned rejection of the proposition that
negative working capital cycles generate a permanent source of cash available
to shareholders and, lastly, (3) correcting the exclusion of capitalized lease
costs from the company’s capital expenditures.
Finally,
to quantify the inaccuracies existing in management’s referenced analyst’s
representations, we will present an analysis that combines Morgan Stanley’s
projected revenue and operating margins with our understanding of how this
performance would flow through to discretionary cash flow available to
shareholders.
Amazon’s Stock Based
Compensation Plan Has A Material, Demonstrated Cost
We
appreciate that this is an old debate, but SBC creates a real cost to
shareholders. Amazon, like most technology companies is typically evaluated on
its Non-GAAP earnings, excluding SBC expense and non-cash acquisition related
charges. For AMZN, however, there is an important twist: Compared to most of
its technology peers, AMZN’s SBC is composed entirely of restricted stock
awards. In contrast to option-based plans, AMZN’s shares need not appreciate
after the grant date for the grantee to realize value. More simply, absent grantee
forfeiture restricted stock units create guaranteed dilution.
Table
one, shown below, demonstrates the historical cost of the company’s ((SBC))
plans. During this reference period, cumulative expenses totaled $4.7 billion
and outstanding shares increased by 47 million. The current market value of those
additional shares is $15.9 billion, indicating that the economic cost to
shareholders has been 3.4 times greater than the recognized ((GAAP)) expense. We
acknowledge that this higher cost reflects the increase in the AMZN’s share
price during this period. The point is that ((DCF)) models that exclude ((SBC))
ignore what is, in effect, deferred compensation that creates temporary cash inflows
to the firm, but that impart real costs to shareholders. To suggest that
restricted stock awards have zero cost is a patently false premise.
Nine Months
|
|||||||||||||||
Year Ended
|
Ended
|
||||||||||||||
TABLE ONE
|
2008
|
2009
|
2010
|
2011
|
2012
|
2013
|
Sep-14
|
||||||||
Stock Based Compensation Expense $((MM))
|
$275
|
$341
|
$424
|
$557
|
$833
|
$1,134
|
$1,089
|
||||||||
Cumulative Stock Based Compensation
|
|
|
|
|
|
|
4,653
|
||||||||
Basic Shares Outstanding MM
|
428
|
444
|
451
|
455
|
454
|
459
|
463
|
||||||||
Change in Basic Shares 1
|
12
|
16
|
7
|
4
|
-1
|
5
|
4
|
||||||||
Cumulative Change in Basic Shares
|
|
|
|
|
|
|
47
|
||||||||
Market Value of Incremental Shares $MM
|
$15,322
|
||||||||||||||
Excess Tax Benefits
|
(159)
|
(105)
|
(259)
|
(62)
|
(429)
|
(78)
|
(121)
|
||||||||
Repurchases of Common Stock
|
100
|
277
|
960
|
||||||||||||
Cumulative Share Repurchases and Tax Benefits
|
|
|
|
|
|
|
124
|
||||||||
Total Cost of Incremental Shares
|
|
|
|
|
|
|
$15,446
|
||||||||
All historical data sourced from Amazon.com filings with the
Securities and Exchange Commission.
|
|||||||||||||||
Market Value and Total Cost data based on AMZN closing price on
December 1, 2014.
|
|||||||||||||||
Negative Working Capital Cycles Create Temporary
Cash To The Firm, Not FCF To Shareholders
AMZN’s
management team states that its financial focus is long-term, sustainable
growth in FCF. We’re good with that. Yet, for a sophisticated group, its
preferred calculation of FCF is simplistic and accompanied by significant,
self-admitted limitations. In its September 30, 2014 10-Q, management cautions
that free cash flow, as they define it presents “limitations due to the fact that it does not represent the residual
cash flow available for discretionary expenditures”. Funny, we thought that FCF
was just that!
With
respect to its negative working capital cycle, basic financial accounting
identifies periodic changes in these accounts as timing differences between
receipt of revenue and payment to vendors. Admittedly, all things equal,
business models with negative working capital cycles are preferable to those
that require sizable investments in inventory. But, their relative value,
particularly in the low interest rate environments is minimal. Rather than
being directly accretive to FCF, their value to shareholders is reflected in
interest income earned on the created float. Importantly, this benefit carries
over to the cash flow statement as a component of net income.
Analyses
that consider net growth in these accounts to be FCF confuse temporary cash
inflows to the firm that should be discounted with FCF accruing to shareholders
that may be capitalized. In AMZN’s case, given the historically rapid growth in
its core retail business, ((CFFO)) has been disproportionately affected this
misconception.
To
put this idea in perspective, let’s consider the following analogy. To us,
counting these timing differences as FCF is analogous to an Amazon customer
concluding that since their package arrives in two days, but their credit card
bill is not due for 30 days that their order was free! It just doesn’t work
that way.
Table
Two below illustrates that, on average, from 2008-2012 these timing differences
represented 42% of AMZN’s CFFO. More recently, as the company’s retail revenue
growth has slowed toward 20%, growth in these accounts has declined, now
representing approximately 14% of CFFO. We conclude this section with the
following observation. Historically, AMZN’s negative working capital cycle has
represented an outsized portion of management’s chosen calculation of FCF. More
importantly, by accepting management’s definition, and consequently
capitalizing this temporary source of cash, the sell-side DCF models that we
have reviewed significantly overvalue AMZN’s equity.
LTM
|
|||||||
Year Ended
|
Ended
|
||||||
TABLE TWO
|
2008
|
2009
|
2010
|
2011
|
2012
|
2013
|
Sep-14
|
Change in Working Capital Accounts ($MM)
|
$714
|
$1,612
|
$1,581
|
$1,464
|
$1,523
|
$767
|
$824
|
Cash From Operations
|
1,697
|
3,293
|
3,495
|
3,903
|
4,180
|
5,475
|
5,705
|
Change in WC/Cash from Operations
|
42.1%
|
49.0%
|
45.2%
|
37.5%
|
36.4%
|
14.0%
|
14.4%
|
All historical data sourced from Amazon.com filings with the
Securities and Exchange Commission.
|
The Combined Influence
of SBC and Negative Working Capital Has Been Staggering!
Table
Three summarizes the combined impact of SBC and the company’s negative working
capital cycle on its CFFO. We do not contest that these two factors positively
impact short-term cash flows into the firm. What we have attempted to highlight
is that they are disproportionately large in AMZN’s case and, more importantly,
that capitalizing them distorts the financial analysis of AMZN’s equity value.
LTM
|
|||||||
Year Ended
|
Ended
|
||||||
TABLE THREE
|
2008
|
2009
|
2010
|
2011
|
2012
|
2013
|
Sep-14
|
Change in WC plus SBC/Cash from Operations
|
58.3%
|
59.3%
|
57.4%
|
51.8%
|
56.4%
|
34.7%
|
33.5%
|
All historical data sourced from Amazon.com filings with the
Securities and Exchange Commission.
|
Leased Based Capital Spending Is Not Accounted
For Under Management’s FCF Definition
Table
four presents AMZN’s capital spending history for both purchased and leased
assets. We have adjusted 2012’s total down to reflect the company’s $1.4
billion real-estate purchase in downtown Seattle. As shown, purchased and
leased capital expenditures have tripled during the past seven quarters, rising
from $3.2 billion to $9.7 billion. As a percentage of LTM sales, this
represents an increase from 5.3% to 11.3%. Coincident with this increase,
AMZN’s management has chosen to rely increasingly on leasing transactions, with
the ratio of leased-to-purchased capital assets rising from 35% during 2012 to
109% during LTM period.
LTM
|
|||||||
Year Ended
|
Ended
|
||||||
TABLE FOUR
|
2008
|
2009
|
2010
|
2011
|
2012
|
2013
|
Sep-14
|
Purchased Capital Expenditures
|
$333
|
$373
|
$979
|
$1,811
|
$2,385
|
$3,444
|
$4,627
|
Capitalized Lease Commitments
|
148
|
147
|
405
|
753
|
802
|
1,867
|
3,901
|
Financing Lease Commitments
|
72
|
188
|
172
|
259
|
29
|
877
|
1,134
|
Total Purchased and Leased Capex
|
$553
|
$708
|
$1,556
|
$2,823
|
$3,216
|
$6,188
|
$9,662
|
Total as a Percentage of LTM Revenues
|
2.9%
|
2.9%
|
4.5%
|
5.9%
|
5.3%
|
8.3%
|
11.3%
|
Year-Over-Year Change in Total
|
76.7%
|
28.0%
|
119.8%
|
81.4%
|
13.9%
|
92.4%
|
42.3%
|
Leased/Purchased Capital Spending
|
66.1%
|
89.8%
|
58.9%
|
55.9%
|
34.8%
|
79.7%
|
108.8%
|
Dep & Amort/Average Net PPE
|
40.2%
|
35.3%
|
30.7%
|
31.7%
|
37.6%
|
36.1%
|
33.7%
|
All historical data sourced from Amazon.com filings with the
Securities and Exchange Commission.
|
Why
does this emphasis on leasing matter? To answer this question, we need to compare
the company’s preferred definition of FCF to the treatment of capitalized
leases under GAAP. To restate, AMZN’s management defines FCF as CFFO less
capital spending. However, under GAAP the capital spending line item is
unaffected by leasing transactions. Consequently, in the company’s preferred definition,
the MS model and the RBC model, absent a small interest component, the cost of
theses leases is ignored. That’s right, $5 billion of acquired assets during the
LTM period that apparently impose no cost to AMZN’s shareholders!
To
correct this false proposition, let’s go back to GAAP. Upon the original lease commitment,
no cash exchange occurs between AMZN and its respective vendors, so rightly,
the company’s cash flow statement does not reflect the value of the acquired
assets as capital spending. However, on the balance sheet, the property, plant
and equipment (PPE) account is debited and long-term liabilities are credited
for the value of the leased assets. As scheduled lease payments are made, the
income statement reflects a non-cash expense allocated to depreciation and cash
interest expense.
Therefore,
under management’s definition, ((FCF)) includes the benefit from the add-back
of the leased asset deprecation. Importantly, the cash portion of the lease
payments reflected as depreciation flows through the financing section of the
cash flow statement. Consequently, it is never subtracted from purported FCF,
essentially creating a free lunch for the company and its street advocates. AMZN’s
business reflects the economic benefit of these assets, yet their cash costs
are never recognized.
Table
Five presents AMZN’s future payment obligations under lease commitments currently
in place, as of its September 30, 2014 10-Q filing. During the next 27 months,
$4 billion of contractual cash outflows are due on these leases. Neither MS nor
RBC reflects these outflows in their respective analyses.
Quarter
|
|||||||
Ending
|
Year Ending
|
||||||
TABLE FIVE
|
14-Dec
|
2015
|
2016
|
2017
|
2018
|
Beyond
|
Total
|
Capital Leases, Including Interest
|
$820
|
$2,006
|
$958
|
$327
|
$135
|
$141
|
$4,387
|
Financing Leases, Including Interest
|
27
|
95
|
96
|
98
|
99
|
943
|
1,358
|
Total Lease Commitments
|
$847
|
$2,101
|
$1,054
|
$425
|
$234
|
$1,084
|
$5,745
|
Data sourced from page 10 of Amazon.com September 30, 2014 10-Q.
|
An Alternative
Presentation of Amazon.com’s Free Cash Flow Available to Shareholders
We
have called into question consensus calculations of AMZN’s FCF. Table Five presents
two separate analyses to clarify our contentions. The first adjusts
management’s calculation of FCF by simply assuming that leased assets were purchased.
This adjustment reveals that in stark contrast to its direct technology and
retail competitors, AMZN has become a significant consumer of capital. This
comparison deteriorates further, as additional adjustments are made to CFFO to
reflect SBC and changes in working capital accounts. As shown at the bottom of the
table, adjusted FCF accruing to AMZN’s shareholders has been negative since
2011. The company’s business opportunity may warrant this level of spending,
but as analysts and investors our first responsibility is to accurately assess a
company’s current financial position. Validly projecting the future depends on
it. Our analysis arrives at a vastly inferior starting point in this process
than does the company, MS or RBC.
We
acknowledge that our analysis makes no attempt to discern maintenance from
growth capital: Our goal is simply to highlight shortcomings in the company’s
preferred calculation of FCF and to caution potential investors to take a
critical look at the line-item components in Wall Street’s DCF models. From our
analysis the economic costs of its various growth initiatives create a risk
profile that is much higher than assumed by its sell-side advocates.
LTM
|
|||||||
Year Ended
|
Ended
|
||||||
TABLE SIX
|
2008
|
2009
|
2010
|
2011
|
2012
|
2013
|
Sep-14
|
Cash from Operations
|
$1,697
|
$3,293
|
$3,495
|
$3,903
|
$4,180
|
$5,475
|
$5,705
|
Less Capital Spending
|
333
|
373
|
979
|
1,811
|
2,385
|
3,444
|
4,627
|
FCF as per Company Presentation
|
$1,364
|
$2,920
|
$2,516
|
$2,092
|
$1,795
|
$2,031
|
$1,078
|
Assume Leased Assets were Purchased
|
220
|
335
|
577
|
1,012
|
831
|
2,744
|
5,035
|
Adjusted FCF as per Company Definition
|
$1,144
|
$2,585
|
$1,939
|
$1,080
|
$964
|
$(713)
|
$(3,957)
|
FCF adjusted for SBC, Working Capital and Capitalized Leases
|
|||||||
Cash from Operations
|
$1,697
|
$3,293
|
$3,495
|
$3,903
|
$4,180
|
$5,475
|
$5,705
|
Less Net SBC Impacts
|
116
|
236
|
165
|
495
|
404
|
1,056
|
968
|
Less Changes in Working Capital
|
714
|
1,612
|
1,581
|
1,464
|
1,523
|
767
|
824
|
Adjusted Cash from Operations
|
$867
|
$1,445
|
$1,749
|
$1,944
|
$2,253
|
$3,652
|
$3,913
|
Less Capital Spending
|
333
|
373
|
979
|
1,811
|
2,385
|
3,444
|
4,627
|
Less Leased Capital Assets
|
220
|
335
|
577
|
1,012
|
831
|
2,744
|
5,035
|
Adjusted Free Cash Flow
|
$314
|
$737
|
$193
|
$(879)
|
$(963)
|
$(2,536)
|
$(5,749)
|
All historical data sourced from Amazon.com filings with the
Securities and Exchange Commission.
|
|||||||
Adjusted FCF figures reflect all direct and deferred costs born
by shareholders. They represent our opinion of the relevant
|
|||||||
financial metrics required to accurately value Amazon.com's
equity.
|
Summary
In
this report, we have recast AMZN’s historical financial results to provide
improved insight into its FCF generation, as well as management’s expanding
risk appetite. To provide context for these adjustments, Table Seven includes data
excerpted from MS’s recently published DCF model. Earlier we noted the heroic
assumptions embedded in consensus thought. As an example, MS projects that adjusted
EBITDA will more than triple over the next four years. As for the model’s
specifics, to the author’s credit, SBC expenses are deducted the equity’s DCF valuation.
However, the two other inaccuracies that we have highlighted both remain: The
author both adds changes in negative working capital balances directly into FCF
and ignores the cost of capitalized leases. We believe that these choices
materially overestimate the projected FCF that is used to value the equity.
To
demonstrate this, we have adjusted MS’s projections by eliminating the benefit
from changes in working capital accounts and deducting the contractual payments
on leases that were highlighted in Table Five. With respect to leases, rather
than projecting new lease commitments to be executed, we have chosen only to
subtract the future lease payments due on AMZN’s existing lease contracts. By
doing so, we implicitly accept the notion that Amazon’s investment cycle has
peaked. We note that doing so defies cautions to the contrary disclosed in the
September 10-Q, where on page 20 management warned that with respect to recent
increases in capital spending that they “expect this trend to continue over time.“
Of
particular note, as depicted in the line Adjusted FCF Used to Value Equity
included in Table Seven, is that from an economic perspective, based on MS’s
EBITDA projections, adjusted to our methodology, AMZN will be a cumulative
negative generator of FCF through 2017.
Lastly,
as a rule-of-thumb valuation metric, the MS report presents FCF yields for each
of the projected years. Interestingly, for this calculation they chose to
add-back SBC and conform specifically to management’s preferred calculation. We
have included our own, distinct, calculation of FCF yield as the last line in
Table Seven. We believe this comparison best supports our submission that
management’s preferred, simplified calculation of FCF, coupled with the
sell-side’s acquiescence to its validity have led to market prices unsupported
by the known facts or projected operating performance. The rewards may come, but
their potential realization entails accepting significantly higher levels of
risk and a longer dated pay-off than is currently acknowledged by the stock’s
proponents.
Year Ended
|
||||||
TABLE SEVEN ($MM)
|
|
2014
|
2015
|
2016
|
2017
|
2018
|
Adjusted EBITDA
|
$6,275
|
$8,494
|
$11,681
|
$15,577
|
$20,723
|
|
Less Cash Taxes
|
(2)
|
(197)
|
(611)
|
(1,422)
|
(2,522)
|
|
Less SBC
|
(1,526)
|
(2,006)
|
(2,515)
|
(3,057)
|
(3,840)
|
|
Less Capital Spending
|
(4,905)
|
(6,337)
|
(7,419)
|
(8,648)
|
(9,950)
|
|
Plus Changes in Working Capital
|
|
1,533
|
1,979
|
1,904
|
2,377
|
2,925
|
FCF Used to Value Equity
|
|
$1,375
|
$1,933
|
$3,040
|
$4,827
|
$7,336
|
Adjusted for
|
||||||
Working Capital
|
$(1,533)
|
$(1,979)
|
$(1,904)
|
$(2,377)
|
$(2,925)
|
|
Lease Payments1
|
|
(1,300)
|
(2,101)
|
(1,054)
|
(425)
|
(234)
|
Adjusted FCF to Value Equity
|
|
$(1,458)
|
$(2,147)
|
$82
|
$2,025
|
$4,177
|
FCF Yield as per Sell-Side Report
|
1.3%
|
2.5%
|
3.6%
|
5.1%
|
7.3%
|
|
FCF Yield as Herein Defined
|
|
0.1%
|
1.3%
|
2.8%
|
||
Sources: MS research dated October 29, 2104 and Amazon.com
filings with the Securities and Exchange Commission.
|
||||||
Adjusted FCF to Value Equity and FCF as Herein Defined are
metrics used by Morgan Stanley, that have been
|
||||||
adjusted to reflect the methodology outlined in this report.
|