Hi Dale Baker; Long mathematical note on the
quarterly revenue series for AMZN, and why I will
be able to cover my short in the single digits
(though I'm more inclined to wait until chapter
11.)
The problem with figuring out the value of AMZN
is that AMZN is a story stock. A story
stock is one that is expected to do great things
at some time in the future, but not in the present.
Story stocks bleed money in the present, but they
have some story that explains why they
should be highly profitable in the future.
The AMZN story is that the current losses are of
no great importance, and that they are instead
building a market. What is important, instead,
is the growth rate in the revenue of the company.
I, and most of the other shorts on this thread,
believe that the growth of AMZN is a lot closer
to being finished than the longs, and that the
recent quarter's results are supportive of our
belief. There has been some suggestion on this
thread that AMZN's recent quarter's results were
not really comparable to the previous quarter's
reseults due to seasonality. The purpose of this
note is to analyze AMZN's revenue numbers with
this caveat in mind. In other words, I am going
to look at the revenue series in ways that will
eliminate seasonality effects.
Raw data is from Wall Street City:
HISTORICAL QUARTERLY RESULTS
REVENUE
(Thousands of U.S. Dollars)
1996 1997 1998 1999
1st Qtr MAR 875 16,005 87,361 293,643
2nd Qtr JUN 2,230 27,855 116,044 314,377*
3rd Qtr SEP 4,173 37,887 153,698
4th Qtr DEC 8,468 66,040 252,893
tscn.com
Seasonality suggests that quarterly results should
only be compared with the same type of quarter,
typically that of the year before. One could suppose
a "strong seasonality" - that the four seasons were
completely independent, and have growth rates that
are incomparable. This might be the case if the
public is celebrating Christmas less and less, for
instance. I will instead assume "weak seasonality"
- that more business tends to be done in certain
quarters, and that the amount of business done in
one quarter tends to be some (fixed) multiple of
the business done in another quarter, all other
things being equal. This is more or less the meaning
of seasonality as used by the Federal Reserve, when
it gives "seasonally adjusted" figures. I will also
assume that the seasonality of AMZN's business has
stayed constant and will remain so. I am hoping for
comments on these assumptions. Hope springs eternal
in the human breast, but don't marry a stock...
AMZN has been growing so quickly that until the most
recent quarter, seasonality has been hidden by the
huge secular revenue growth rate. The AMZN longs
would have us look only at the year over year
figures, which are still great, but this puts an
investor into a precarious position. Under this
restriction, if AMZN quit growing right this quarter,
we wouldn't be able to detect it until about a year
from now. In the event that other investors were not
so stupid, we would end up in a situation where we
might have to sell our shares after everybody
else... Somebody recently posted a comment about
stock that I liked: The first rule in investing
is don't panic. The second rule, is that if you
must panic, do it before everybody else. So
what I am doing here is showing a way of looking
at the quarterly numbers in such a way as to detect
a reduction in the growth rate without having to
compare consecutive quarter's growth rate, but also
without having to wait for the full year data lag.
The first thing to do is to convert the above raw data
into a table of data that shows quarter over previous
year's quarter growth rates. This is what the longs
would have you do with this data, but on a company
who's growth rate is rapidly dropping, this estimate
of the growth rate will be way too high. This is
because the year over year figures include 4 quarters
of growth rate, only one of which is the latest and
most cogent.
I've added two columns of calculated numbers to the
usual Yearly Increase figures:
AMZN Growth Rates:
Yearly Growth
Date Revenue Increase Change Lossage
---- -------- -------- ------ -------
Q96 875
2Q96 2230
3Q96 4173
4Q96 8468
1Q97 16005 1729%
2Q97 27855 1150% -579% -33%
3Q97 37887 808% -342% -30%
4Q97 66040 680% -128% -16%
1Q98 87361 446% -234% -34%
2Q98 116044 317% -129% -29%
3Q98 153698 306% -11% -3%
4Q98 252893 283% -23% -8%
1Q99 293643 236% -47% -17%
2Q99 314377 171% -65% -28%
In the above table, the Yearly increase is
the revenue rate of increase from the same quarter,
previous year. These figures, since they compare
similar quarters, are seasonally adjusted. If
seasonality consists of certain quarters having a
larger proportion of business than others, than that
ratio will divide out. The resulting series labelled
Yearly Increase are therefore seasonality free,
and can be manipulated on a quarter to quarter basis.
The fact that these Yearly Increase numbers
form a beautifully smooth series is an indication
that the seasonality has been removed from them.
The first manipulation is to take the differences
between consecutive Yearly Increase rates.
The Change column does just that. The Change
figures have a secular tendency to decrease considerably
in magnitude, and that means that they are hard to
estimate for future quarters. A nicer data series is
the Growth Lossage, which is the percentage
change in the Yearly Increase taken from
consecutive quarters.
As an aid to those wishing to put all this into a
spread sheet, so they can modify the assumptions
and compute their own growth rates (just like those
highly paid securities analysts who are currently
bailing out of this stock), here are some sample
calculations for 2Q99:
Yearly Increase: ((314377 / 116044) - 1) * 100 = 171
Change: 171 - 236 = -65
Growth Lossage: (65 / 236) * 100 = 28
Examining the above table, it is perhaps surprising
that AMZN's year over year growth rate has done
nothing but decrease. That is, all the entries in
the Change column are negative. Another way
of putting this is to say that AMZN's growth rate has
been decreasing - they are growing slower with each
passing quarter.
The second thing to notice is that between 2Q97 and
3Q98, AMZN's rate of growth loss was decreasing. That
is, AMZN seemed to be coming closer and closer to
maintaining a high growth rate. This peaked in 3Q98,
which had a 306% growth rate, just a little less than
2Q98's 317% growth rate.
The final thing to notice is that in 4Q98 between 2Q99,
AMZN's loss in growth rate has again accelerated. By
looking at the Growth Lossage numbers, it is
clear that what actually happened is that for 3Q98 to
4Q98 AMZN had an unusually low loss in growth rate.
I believe that this is due to the high level of
advertising done by AMZN during those quarters. With
the recent quarter's results, the return to a
historical loss in growth rate of around 23% per
quarter is clear.
Now that we have a reasonable handle on the revenue
series, we can do some extrapolation into the future,
and get some estimates as to what AMZN is really
worth today...
AMZN Growth Rates 3Q99 &c. extrapolated:
Yearly Growth
Date Revenue Increase Change Lossage
---- -------- -------- ------ -------
Q96 875
2Q96 2230
3Q96 4173
4Q96 8468
1Q97 16005 1729%
2Q97 27855 1150% -579% -33%
3Q97 37887 808% -342% -30%
4Q97 66040 680% -128% -16%
1Q98 87361 446% -234% -34%
2Q98 116044 317% -129% -29%
3Q98 153698 306% -11% -3%
4Q98 252893 283% -23% -8%
1Q99 293643 236% -47% -17%
2Q99 314377 171% -65% -28%
3Q99 356579 132% -39% -23%
4Q99 508314 101% -31% -23%
1Q00 522684 78% -23% -23%
2Q00 503003 60% -18% -23%
3Q00 520605 46% -14% -23%
4Q00 705623 35% -11% -23%
So by 4Q00, I am looking at a company with a year
over year growth rate of around 35%, and sales of
something like $2.25B ttm. This is a little over
twice AMZN's current ttm sales.
AMZN is a retailer, and the industry average profit
margin (according to the above link) is 3.21%. I
really don't think sales of $2.25B will be enough to
make AMZN profitable, particularly since the trend
in profitability (which I will do an analysis of later)
has turned so amazingly negative, but if they
could achieve this level of profitability it would be
$72MM per year. Given a revenue growth rate, at that
time, of 35% per year, the stock market might put a
multiple on these profits of 35, for a market cap of
$2.5B. The current shares outstanding number is 162MM,
but AMZN is leaking more shares out at a rate of
something like 11% per year, so I will assume 180MM
shares. This gives a stock value of $2.5B/180MM =
$14 per share. Of course there is going to be a 2
for 1 split in a few months, so the actual stock
value would be likely in single digits. Remember
that 18 months is not very long from now.
Given a value of $14 per share 18 months in the future,
the current value would be somewhat less, depend on
the risk adjusted rate of return. Given AMZN's
incredibly high volatility, (and losses) I would be
amazed if people were buying this stock with an
expectation of a return of less than 20% per year.
Eighteen months of 20% to get to $14 gives a current
"fair" value of AMZN of around $10 5/8. Of course
the upcoming split takes this value well into single
digits.
The other question is whether or not AMZN will be
able to post profits with sales of only about twice
their current annual sales. I suggest not, but
will do more analysis of this later.
-- Carl
P.S. If you wish to risk $375 going short AMZN, and you
are worried about, say 30 point moves going against you,
you can always short 10 shares. |
