If you've held shares of 3D Systems (DDD 3.45%) this year, you've watched the share price get cut in half from its high at the beginning of the year. If, however, you followed Citron Research's (and others') advice to short the shares, you've made out pretty well.
Should you have followed Citron? Or was Citron just lucky? After all, it was wrong with the timing of its 2013 short report, with shares climbing 140% between February 2013 and the beginning of 2014. In its early 2014 report, Citron doubled down on its short thesis, but the report suffers from two glaring errors that could have cost investors.
Second time at bat
In the previous article in this deep-dive look into 3D Systems, I found the 2013 short report by Citron to be full of hyperbole alongside statements that Citron couldn't be bothered to prove why the bull case was wrong. As a scientist by training, I was bothered by that lack of data.
Citron's January 2014 report on 3D Systems, however, was a better read. It contained less hype, fewer outrageous statements, and more substance. Comparing companies in the same industry on a price-to-sales multiple is great. Looking at historical short interest is super. Printer sales growth at the company and its competitor, Glassdoor ratings, product reviews -- all those data points are awesome, and they make me wish Citron had taken this approach originally.
Unfortunately, Citron cannot seem to avoid meaningless barbs, which detract from the investment case it is trying to make. For example: "Great, so now along with the Justin Bieber Vibrator [referring to the 2013 report] you can now get your custom Star Wars sex toys." How does this help Citron's argument?
Where Citron goes wrong
Rather than going point by point through this report, I'll try to discuss several of its worst failings. Citron's arguments are in italics, and my responses follow.
Since "the premier expert in the industry" said that 3D Systems is overvalued, Citron believes it must be so.
Actually, most experts are pretty bad at predicting the future. James Montier of investment firm GMO cites research showing so in Seven Sins of Fund Management. Feedback on prediction results is key in determining how much one should trust expert predictions. Illustrating this point, Montier points out that weather forecasters are quite accurate probably because feedback is so quick and emphatic. Stock price predictions from experts, on the other hand? Worse than a coin flip.
Citron presents a pie chart showing market share, highlighting 3D Systems and Stratasys, and then confuses industry segment market share with total industry market share.
This interesting chart from Terry Wohlers, the previously touted expert, doesn't look good for 3D Systems. Assuming he's accurate, Stratasys (SSYS 1.63%) has 53.4% of the installed 2012 industrial 3-D printer systems market share, measured by units (adding together Stratasys and the since-purchased Objet), compared with 17.5% for 3D Systems. This almost certainly reflects that unlike Stratasys, 3D Systems has not emphasized the industrial market in the past, preferring to build its presence in the consumer market.
However, the table above the pie chart is misleading. It implies that the market share numbers in the chart are total 3-D printing market share, rather than just industrial 3-D printing market share (which is how the pie chart is labeled). Citron should either adjust the revenue numbers it supplies for 3D Systems and Stratasys to reflect only industrial revenue, or it should make clear that it is assuming the market share for industrial 3-D printing for the two companies was the same as for the total markets they serve.
By not doing either, it misleads investors reading the report into believing that one set of numbers applies everywhere, when in fact it may (and probably does) not.
Citron attempts to value 3D Systems as if it were Stratasys.
At the end of Page 11, Citron does something I found really bizarre. It says that because Stratasys has a larger market share (in industrial 3-D printing, but remember that Citron is using a broad brush here) and better products and higher likely sales (based on some data presented earlier), investors should assume that Stratasys and 3D Systems should have the same enterprise value. Investors should then add 3D Systems's net cash to get market cap and divide by the share count to get an "indicated" share price of $55.40.
I've never seen that done before. Using relative valuation, analysts would normally use the industry's average of some ratio such as enterprise value-to-sales, or EV/S, as a "fair" ratio to apply to 3D Systems, and then project what the market cap would be from that. This corrects for different absolute sales levels.
If we use Stratasys' EV/S ratio instead of the industry average, however, the end result of a $55.42 share price is pretty much the same because, by a quirk of analyst estimates, the two companies coincidentally have essentially the same projected 2014 revenue. If analysts had projected different total revenue for 3D Systems and Stratasys -- say, $900 million and $600 million -- the projected 3D Systems share price would have been quite different.
The calculations behind Citron's projected fair value of $46 per share are wrong on two counts.
Citron projects 2021 earnings for 3D Systems by setting up a bunch of assumptions. Reading beyond the hyperbolic language (e.g., "let's pretend 3D Systems is able to accomplish the impossible"), these boil down to:
- Let 3D Systems grow at the same 19% annual growth rate of the industry, using numbers from Terry Wohlers, where the industry is projected to grow from $2.2 billion in 2012 to $10.8 billion in 2021.
- "Round up" to a 20% market share, even though the company has only 17.5% (but remember, that's only for industrial 3D printing, even though Citron is applying this to the entire 3-D market).
- Give the company a 20% net margin.
- Let the company avoid paying any taxes (which doesn't make sense, as net margin is after taxes).
- Allow no share dilution.
(Using the preceding industry growth numbers, by the way, along with a projected 2014 revenue given a couple of pages earlier, 3D Systems actually has a 21.6% projected market share in 2014, not the 20% level Citron gives the company, nor the 17.5% level from that pie chart, which Citron claims is the real total market share.)
From those assumptions, Citron projects $2 billion in revenue in 2021, "roughly $400 million" in net income, and $3.67 EPS in 2021 -- which, by the way, assumes 109 million shares, not the 102.78 million shares Citron earlier said the company had. Citron then discounts its 2021 EPS figure back to the present and assigns a 20-times P/E multiple to get a share price of $46, at a time when 3D Systems shares were trading at $79.87.
There are two problems with this. First, the math doesn't work out.
Using the 2021 market size of $10.8 billion, a 20% market share, and a 20% net margin, and the previously supplied number of shares, this works out to $4.20 EPS in 2021. Doing the same discounting and applying the same P/E multiple results in a share price of $52.74.
Second, Citron combined two different modeling methods.
Citron discounts the 2021 EPS number back to the present. However, in a discounted cash flow model, or DCF, the value of a company is the sum of all the future cash flows discounted back to the present using a suitable discount rate. Citron discounted a single year of cash flow -- 2021's earnings -- which ignores all the cash flows except for 2021. You cannot do that in any sort of DCF model, unless your argument is that 2021 is the one and only year in which 3D Systems will ever have earnings. (Not even Citron claims that.)
Furthermore, Citron then applies an earnings multiple to the present value of 2021's earnings. That's fine, except it did this at the wrong point in time. By applying an earnings multiple, the analyst assumes that all future cash flows from that point forward discounted back to that point are worth whatever the multiple indicates (here, 20 times this single number). This is one way to calculate the terminal value portion of a DCF model, but this is done at the end of the explicitly modeled period (e.g., 2021), not the beginning.
Instead, Citron should have done one of two things:
- Modeled EPS for each year between 2013 and 2021, and used the 20 P/E multiple to calculate a terminal value as of 2021. It would then discount all of those EPS values and the terminal value back to the present to get a valuation today.
- Looked at EPS at some point in the future, which it did, and then applied the multiple on it at that point in time and said, "This is the projected stock price in year X." If Citron did that, it'd have $73.40; using my math, you'd get $84.
You could then compare that future share price with the current price ($79.87 on the report's publication date) and make a judgment about the return between those two share prices. But $73.40 in the future versus $79.87 now? No, thanks; obviously overvalued today. Using my numbers: $84 in the future versus $79.87 now? Just 5% growth total over eight years. No, thanks; also overvalued today.
A few strong arguments can't cancel out the weak ones
Despite these mistakes, this report does raise some good, potential issues:
- Industrial 3-D printing is (currently at least) very important, and 3D Systems is weak there.
- There are potential issues with customer service at 3D Systems.
- A company that's been priced as a growth company cannot afford to fail to grow.
- 3D Systems could be having difficulty squeezing synergies out of the many acquisitions it's been making.
- The 3-D printing sector includes a lot of competition.
However, Citron's two glaring mistakes -- misusing DCF valuation and the (deliberate or not) confusion between industrial 3-D printing market share and total 3-D printing market share -- make it very tempting to discard their research out of hand. And I feel that the flamboyant, belittling language (e.g., "Now we wave our magic wand and grant them an over-the-rainbow 20% net income margin") doesn't make the decision any easier.
If you want to follow Citron in shorting other companies, I recommend you read its reports with a critical eye even if the reports contain useful information. If the analyst is making similar mistakes, especially in valuation, that could end up costing you.
Readers can find each article in the series by clicking here.
