It's weird, the Russell is down 25-30% from its June peak, so some broad bad thing must be going on, and CEOs usually aren't shy about blaming external factors - "it's not market share loss just orders being pushed out" - but none of my favorite companies are complaining like that. The ones with both internal and external momentum, which is most of them, are cheaper than they were nine months ago because their shares have merely treaded water. And the prices of the handful with serious internal problems have been obliterated.
I've been meaning to write something that summarizes all my favorite ideas in one place. Each idea has something in common - industry or style wise - with at least one other so I thought it'd be efficient and fun to try to provide a feel for each integrated in a single report.
I had been putting it off but the sell-off sparked urgency: my opportunity set is the best it's been since 2010 and I want these predictions on the record. So here goes.
You know what categorization is: the world is messy, no two things are exactly alike, but it makes life easier to lump similar things under a common label. That way when we want to talk about a cat we can just say "cat," calling to the listener's mind the great many features that all cats have in common.
In biology, it's simple to find things with common features to lump together because evolution has a mechanism that segregates all the messiness into distinct categories - speciation. But business isn't like that. It's messier and yet we still categorize of course. But it's harder to do.
Category errors occur when, a) a thing is different from other things in its category on some important features, and b) people don't notice that difference. Time for an example.
What's a contract manufacturer (CM)? A firm that makes what another designs. That's true for all CMs. Usually, what they make is electronics, which is a $100b+ industry, where outsourcing has been utilized for decades. As a result, most CMs are alike in these other ways: many nearly identical competitors, negligible market share and pricing power, low margins, low ROIC potential, capital hungry, and low growth.
Since CM performance clusters in a small unimpressive range, the CM category is valued cheaply and uniformly: 8-13 P/Es are typical.
Fabrinet (NYSE:FN) is functionally a CM - it manufactures what others design. It's valued like a CM - cash adjusted P/E of ~10. But economically, it sure doesn't look like a CM - its ROIC is consistently 20-30% driven by elite margins and low capital intensity. And it's grown fast without acquisitions.
Fabrinet specializes not in electronics ($100b+) but optical components, which is <1% as large and much younger. They were founded in 2000 to rollup the first plants and programs being vertically disintegrated by the industry's OEMs. It's the only pure play optics CM in the world, and likely manufactures ~50% of what's outsourced in the industry. Their elite economics follow from those facts, as I've been explaining.
And yet sell-siders treat FN like a plain old CM, lumped into the same data tables as all the others, slapped with the same low P/E, justified by the sort of crappy features associated with electronics CMs in general that simply do not apply to Fabrinet.
Speaking of category errors, Farmer Mac (NYSE:AGM) does in agriculture what Fannie Mae does in residential housing. Mostly. What they've got in common is access to the cheap finance that comes when lenders think the government has your back, which lets you profitably hold low risk low spread loans purchased from 3rd party originators. Throw in extremely low overhead and extreme leverage privileges and you can transform assets that yield 1-3.5% into a heck of an ROE.
Now, when I say they're like Fannie, the question is "which Fannie?" The one that ran this model competently in the 1980s and 1990s and was one of the best performing stocks in the universe? Or post Y2K Fannie that essentially obliterated itself?
The key to sustainably milking these privileges is never taking credit risk. Honestly, that's not that hard to do if you're committed to it; 20% down on a properly appraised property purchased by a married guy with a job and a decent credit score does the trick. Fannie imploded because it strayed from that for various reasons, i.e. to get back market share from maniacal private competition unleashed by private label securitization, and to make the Clinton's and Bush's happy by meeting aggressively stupid affordable housing goals.
Farmer Mac won't stray. The risk here is, ironically, that agricultural borrowers are too sane. They've used good times to pay down debt, and farming doesn't attract many get-rich-quick maniacs that are willing to forge documents to get that corn farm in Iowa. Also, there's another GSE in the farm space called the Farm Credit System - too complicated to describe here - but it's got similar advantages to AGM and a similar mandate.
So the risk is AGM is redundant. Along those lines their spreads have narrowed in recent years, and their ROE is down to 13-15% (~$4-4.50 EPS on $30 book value), much less than Fannie achieved in good times (mostly because Farmer Mac's leverage is about half as high). But spreads have stabilized, AGM's core business is growing, and they've got some other neat ways to generate ultra low risk portfolio assets (lending to Rural Utilities for example).
At 1x book and well under 10x EPS, I like it. It's priced the way it is because, as I was repeatedly warned by commenters in my AGM writeup, all GSEs are corrupt, reckless, doomed, and probably evil too.
Back to contract manufacturers - here's an electronics one called Key Tronic (NASDAQ:KTCC). They're not as structurally advantaged as FN but still quite unique in two ways. First, most manufacturers their size ($500m revenue) have limited internal capabilities. They'll solder stuff on circuit boards for you, test it, etc., but if you want them to build a complete product containing plastics and metals, they'll have to source that stuff from a 3rd party.
OEMs don't like that - long chains break more often than short ones. So small manufacturers mostly just do circuit boards, which is too bad because making parts is a less sticky thing (less fixed cost and time in setting the thing up) than are complete products. Big manufacturers have broad capabilities but they're not interested in your stupid $10m program, so OEMs with low volume products have limited options. KTCC is an excellent one - run by excellent long-term managers - and that's why they're winning so many new programs right now, and why I expect fast sustained organic growth.
The second neat thing is that their prime plant is in Mexico, where wage growth compared to other low-cost manufacturing centers like China has been lower by 5%+ per year for more than a decade. The cost calculus has shifted and of course Mexico was always more geographically and linguistically convenient and less willing/able to steal your IP, so big "boomerang" accounts are flying back from China.
Program churn in the form of big legacy customers overpromising on volumes and the costs of ramping so many new wins has hurt KTCC's margins temporarily. They're near 2% operating right now, which translates into EPS of ~$0.50 annualized. Each 100 bps of improvement increments EPS by ~$0.30. So a return to the 3-5.5% range KTCC usually achieves when things are smoother implies EPS of $0.80-1.25. Throw in 10-20% revenue growth and the EPS potential 12-24 months from now is huge compared to the $7.50 stock price.
eMagin (NYSEMKT:EMAN) is my lottery-ticket. They make microdisplays - picture HD TVs hit by Rick Moranis' shrink-ray, a useful size if you want to strap the display to your head. And theirs are OLED - that's organic light emitting diode, which translates roughly as "tiny light emitting ~transistors made of carbon plus some other stuff," which are set up in a pixel array. The common non-OLED alternative to emitting light at each individual pixel is to filter it there instead. Filter what? Luminance from a backlight, a bulky energy hungry thing OLED doesn't need. OLED is the future (at least for a while).
Bear with me as we zoom forward: OLED is very hard, a double-edged sword, big learning curves limit competition, often atrocious yields and high costs, volatile gross margin, most needed equipment entirely custom built (long lead time and uncertain performance); high cost/price displays so 75% of EMAN's ~$30m revenue is military, which continues to fund EMAN R&D ~designed to help them monopolize US military applications and displace only military rival back-light filter based Kopin (NASDAQ:KOPN); military/high-end quasi-monopoly alone justifies legit valuation but a couple of years of financial regress obliterated hope; current market cap <$50m.
OK, back to regular speed. EMAN sells 25-75k units per year today; consumer AR/VR is set to go, VR in particular, and successful individual devices are likely to sell millions of units per year. EMAN's displays are currently too expensive, but it's a chicken-egg problem - to drive down cost you need volume and to get volume you must convince others that cost can be conquered. Hard to prove but EMAN's got someone intrigued because just a few weeks ago, an unnamed tier-1 VR developer licensed the EMAN built VR-prototype they made to show off what their displays made possible [edited 2/3: The presser did not in fact indicate anything about the partner's identity. I mistakenly thought it did because during the Q3 CC management spoke at length about tier-1 VR companies that they were in talks with. My mistake.]
Whereas eMagin could triple between 4 PM one day and 9:30 AM the next, Simulations Plus (NASDAQ:SLP) is more of a slowburn 10-year 10-bagger candidate. I'm not saying that's likely just possible.
The idea is to develop drugs by using computers to predict properties of never before synthesized molecules. That's doable because structurally similar molecules (size, shape, weight, how many nitrogens) have similar biological properties (absorption rate, membrane permeability, etc.). So from an ever-expanding database of already-tested molecules, we can ever more accurately search for that needle in the haystack of physically possible but never before synthesized molecules, using computational tricks (neural networks) that began advancing rapidly in the 1990s.
Actually, 2/3s of SLP's software revenue comes from a type that operates later in the drug development process, making mechanistic (not a neural net) predictions that help to refine already chosen molecules. It's a near-monopoly product, still growing fast, widely used, and is the major reason they've been ferociously profitable for years.
But it's the neural net - the needle in a haystack searcher - that's most exciting. SLP is a leader in biosimulation yet their revenue is ~$20m/year. How's the leader so small in a field, pharmaceuticals, where $20m is a rounding error? Because until quite recently, biology was too complex for computer modeling to add much value. SLP's CEO/Founder Walt Woltosz - who invented the thing Stephen Hawking uses to communicate - talks all the time about how far behind the field it's compared to aerospace simulation where he spent his early years.
Think of all the features a molecule must have to be a useful drug - it's gotta absorb into the blood, attack the right proteins but not the wrong ones, the body must be able to break it down and piss it out, etc.
As software improves, the fraction of drug features it can predict accurately rises. SLP leads and yet is still small because that fraction is still quite low. What will the biosimulation leader's revenue and market cap be in 2030? SLP's current market cap is $180m.
In complicated situations, I like to have something I can hang my hat on; a thing that's almost certainly true, valuable, and downside limiting. For eMagin, it's their military dominance; for SLP, it's that 95%+ of their licenses (measured by dollar value) renew each year. For Nova Measuring (NASDAQ:NVMI), its leanness.
Nova has got a doppelgänger called Nanometrics (NASDAQ:NANO): same size, almost precisely overlapping product lines and customers, etc. But Nova spends about half as much on SG&A. The overhead gap alone means a 10-15 point advantage in operating margin; the actual gap has tended to be slightly wider since Nova's gross margins are higher.
Measured over any horizon longer than a few quarters, Nova has also grown faster and gained share vs. Nano, and they've got the cash to bolt-on complementary companies, but despite all this on P/S and P/B, the firms are equally valued. It's insane.
Nova is an Israeli company that makes the tools used to see microscopic chip features to make sure each manufacturing step is going smoothly. "See" in this context means bouncing short-wave rays off a wafer's surface to generate data which software processes into a rough picture of what's going on, i.e. are transistors spaced by 7 nm +/- some tiny tolerance.
Only a few chip companies participate at the cutting-edge and Nova does extensive business with all of them. TSMC (NYSE:TSM), their most important customer, retrenched in 2015 which Nova more than offset with wins at Samsung (OTC:SSNLF) and others. TSMC told us the other day their capex will snap back in 2016, so I think the year is going to beat what analysts expect. Even if I'm wrong, Nova is cheap at $9 with a cash-adjusted EV under $7 and a $0.70 run-rate.
Israeli tech companies do things lean apparently - and Mr. Market doesn't seem to care. Here's one - Silicom (NASDAQ:SILC) - with 20% operating margins that'll earn ~$2.20 (per analysts) or $2.50+ (my opinion) in 2016, yet its stock price is $30 and its cash-adjusted EV is just $22. Like Nova, they've grown rapidly in the last decade, hiccupped a bit recently, and have paid for it excessively.
SILC makes server cards, little circuits packaged in plastic with some I/O and SILC-designed software customized for a customer's special need, usually security related, but often with speed or power/bit conservation considerations too. The idea is to add capability to a generic server with a simple plug and play device.
A key aspect of the market is that it's high mix low volume and rapidly changing. SILC's ~$100m run-rate consists of a few sizable SKUs and customers, and a hundred plus small ones. Chip big boys mostly aren't interested in competing here, especially since the overall size of the market is limited. So little competitor effort flows top down. And they're protected from upstarts because of experience and reputation with existing customers.
Like SILC, Sierra Wireless (NASDAQ:SWIR) also makes little computational modules that do specific, narrow things. For SILC, security is #1; for SWIR, it's wireless connectivity. They make about 1/3 of the wireless cards embedded in machines connected to the internet, i.e. cars, meters, and industrial machinery. This market is much larger than SILC's and therefore more competitive. Whereas SILC does 40%/20% margins, SWIR does 30%/5%. Notice that SWIR's overhead is higher than SILC's. Why? Both sell globally, but only SWIR has global physical presence.
For whatever reason, local marketing and support is required to sell in M2M. So whereas both these companies see long growth runways and invest accordingly, the land-grab has been costlier to pursue in SWIR's case. While both companies should keep growing 10-15% per year organically, SWIR has more leverage and I expect to see margins trickle up towards 10% at a rate of 100-200 bps per year. Since SWIR's $480 market cap is 70% of $680m in forward sales, so if I'm right on growth and margins, this will be a big winner.
So that's it for the company summaries. Again, the goal was to communicate a feel that'd transfer some momentum to anyone interested in a deeper dive. The summaries were light on numbers so I'll finish up with some numbers, specifically a table containing my best guess (a gambling over/under not an intrinsic value estimate) for where each stock trades at the end of 2016 and 2018, and also a range, a sort of 80-90% confidence interval. Below the table, I briefly explain my guesses.
Stock Price Over/Unders For End 2016 and End 2018
|Price 2/1/16||End 2016||End 2018|
|Ticker||~Actual price on 2/1||Best guess, (lo-hi)||Best guess, (lo-hi)|
|FN||$24.50||$30, ($20-35)||$40, ($20-50)|
|AGM||$32.50||$35, ($28-40)||$50, ($30-70)|
|KTCC||$7.50||$10, ($7-12)||$15, ($7-20)|
|EMAN||$1.40||$2.25, ($1.25-3)||$5, ($0.75-10)|
|SLP||$11||$12.50, ($9-15)||$20, ($11-25)|
|NVMI||$9||$12, ($8-15)||$18, ($7-25)|
|SILC||$30||$40, ($28-45)||$50, ($28-65)|
|SWIR||$14||$18, ($12-22)||$25, ($14-35)|
[FN reported after market and will open up on 2/2. I've left everything unchanged including my 2016 and 2018 FN estimates - and the verbal justification - which would otherwise be revised higher slightly. I guess you'll have to take my word for all this.]
- FN is cheap, has lots of operating momentum, and the stock is trending up. Hard to see 2016 going poorly - easy to imagine $30 at year end, which implies a P/E of ~17, or more like 13 adjusted for cash. Long-run downside is protected by legitimate competitive advantage, but is capped somewhat by the gradual nature of growth and limited operating leverage in contract manufacturing.
- I've been waiting a while for AGM to get revalued up from this 7 P/E non-sense. I don't know when it'll happen, but consistent 13-18% ROEs will eventually loom larger in investor minds that the tar of the GSE label, especially as memories of 2008 fade. That's why I've got more confidence in AGM over three years than one.
- At some point, probably starting now, KTCC will string together 3-4 quarters of sequential growth in revenue and margins that'll get EPS back above $1 and the stock back above $10. Once that happens, further upside will depend proving the story re Mexico and capabilities/size driving superior organic growth and margins.
- eMagin is the primary reason for including the price range predictions. Am I sure EMAN is going to grow outside the military domain? No. But if it happens and 50k units turns into 5m within the span of 18 months, look out. And the odds of something like that happening are much higher than is priced. Plenty of cash and military dominance mean it can't go to zero.
- In the short run, SLP's price will be driven by P/E movements - hard to predict and can't count on expansion at a 35 P/E even though they're growing fast. But in the long run it's about how big this market gets and whether SLP can maximize the value of its leadership - I'm optimistic on both fronts and can imagine holding SLP for a decade or two.
- NVMI is cheap today ($7 EV, $0.70 run-rate, ~15% margins, fat ROIC), growing, highly profitable, incredibly lean, it prospered in 2015 despite TSMC retrenchment because they broadened their customer base, and now TSMC is coming back. Hard to see how you lose money here; they'll grow profitably, and deserve a much higher multiple. I can't predict when that comes but it will eventually if they keep it up.
- Like NVMI, SILC has proved itself with superior growth and profitability for many years. It stubbed its toe recently but just posted an incredible quarter and appears to have its momentum back. How long can an EV/EPS ~10 persist for one with 20% margins with a strong competitive position in a fast growing IT niche?
- I expect relatively smooth progress from SWIR: 8-15% organic growth per year, some bolt on acquisitions, and gradual margin expansion from 5% to 10% over several years. And I expect multiples to sway wildly during that process. Sierra's P/S multiple is ~0.80 right now. That can't last if I'm right about the fundamentals.
Disclosure: I am/we are long FN.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: I am long every stock favorably mentioned in the piece.
Editor's Note: This article covers one or more microcap stocks. Please be aware of the risks associated with these stocks.