Morgan Stanley’s Amazon.com EBook Analysis Is Seriously Flawed
On January 6th, 2013 Morgan Stanley analyst Scott Devitt upgraded Amazon.com to Overweight on the thesis the company would gain e-Commerce sales market-share and grow revenue by around 30% annually in the next 3 years (2013 +31.2% y/y, 2014 +30.3% y/y, 2015 +27.3% y/y).
This bull thesis went out the window weeks later when Amazon.com reported Q4-2012 sales growth of only 22% y/y missing the average sell-side sales estimate by $1 billion with an epic 700bps ex-FX y/y sales deceleration from Q3-2012. The Morgan Stanley analyst quietly lower his sales estimate by 1/3 to around 20% y/y annual sales growth over the next 3 years after the report.
Yesterday the same Morgan Stanley analyst came up with a new bull thesis. In a new sell-side note citing a U.S. consumer survey they did of 1108 owners of eReaders and tablets, Scott Devitt concluded the 2012 eBook market is 50% larger than they previous estimated.
Using his assumptions of EBooks sold per hardware device tie-ratios, e-Reader/tablet hardware market growth, and market-share, his Amazon.com’s eBook operating profit estimate ballooned higher 3 years out.
Investors giddily bid up Amazon.com shares by +4.16% on Morgan Stanley’s revised higher EBook market analysis. The problem is his methodology is seriously flawed.
Morgan Stanley asked e-reader and tablet owners how many EBooks they bought per month. Then they multiplied the number by 12 and took a "finger in the air" 40% discount due to "annualizing" it and also guessing there probably is a lower International tie ratio.
The end calculated results were 13.3 EBooks purchased per e-reader device and 6.4 EBooks purchased per tablet in the year of 2012. He then used these same ball-park tie ratios as a base case along with device growth to extrapolate EBook market-size and Amazon EBook revenue & profit over the next few years.
If Morgan Stanley was going to annualize the EBook purchased per device number, why didn’t they just ask how many EBooks each consumer bought in whole year of 2012 instead of per month? Wouldn’t a consumer who actually bought 0.25 EBooks per month likely to say 1 instead of 0? Doesn’t this introduce large rounding error issues? Isn’t this a serious flaw?
What’s with the made-up 40% discount number which comes arbitrarily out of thin air?
Why would tie ratios stay in the same range 13-14 Ebooks a year per e-reader device and 6-7 Ebooks a year per tablet device over the next 3 years? Wouldn’t the tie ratio change as the market matures?
Also the final tie ratios of 13.3 and 6.4 EBooks sold annually per each and every hardware device didn’t pass the smell test to me either, so I did a survey of my readership.
I actually asked the right question of how many EBooks they bought in the entire 2012 year instead of the errant per month method Morgan Stanley used. I got 85 responses for my survey. The 2012 EBooks bought per eReader device tie ratio came out to 4.8. The 2012 EBooks bought per tablet device tie ratio came out to 1.4.
My reader base demographic is probably more affluent and more educated than the average American and E-reader/tablet owner. The tie ratios are LESS THAN HALF AND A FRACTION of Morgan Stanley’s number, who used less accurate survey methodology.
Morgan Stanley’s assumptions and methodology garbage in = garbage out, just like Scott Devitt’s original Amazon.com 30% annual sales growth bull thesis. Investors beware.