Description

We believe Celestica is one of the cheapest stocks in the market today, with very limited downside and meaningful upside.  It is an unlevered business trading at approximately 5.9x earnings ex-cash, an EV/EBITDA ratio of about 3.5x, and at an unlevered FCF yield of 20%.  The business has a combination of cash and “hidden assets” that, when taken together, amount to about $3.60 share, or nearly 35% of its market cap.  The company has retired 38% of its shares in the last 7 years and recently authorized another buyback of 10% of their shares.  CLS trades at about a 3x ebitda discount to peers despite being a comparable business to all of them.

There is little debate that Celestica is cheap on trailing numbers or relative to peers – there is a bit more debate about how good a business this is and its future earnings power.  We believe this is an intensively competitive, but above average business, as demonstrated by its very consistent cash flows and a teens after-tax ROIC over the last 10 years.  We also believe investors have unfairly stereotyped Celestica as being less good than its peers, as evidenced by a historical discount on an EV/EBITDA basis of about 1 turn of EBITDA, which has recently widened to 3x.  We think growth prospects are solid and the business has several “hidden gems” in it that the market doesn’t appreciate.  In general, we also think more highly of the EMS industry in terms of business quality than others (though this isn’t saying much).

Even if you believe traditional EMS is terrible, and the structurally declining parts of the business are worth zero, we think the stock is worth $13.50 just on the basis of its good and growing business, which we believe is superior to Plexus.  And given that the declining business generates a lot of cash, has a teens ROIC, and has nearly $400M of working capital, its value should be something greater than zero.   If we value the legacy business just at working capital and the growing business in line with Plexus, the stock would trade at $16.  A more traditional approach, just valuing the biz at 5.5x ebitda, would result in a $15 price.  We see downside around tangible book value, which is about $9 a share.  At its current price of $10.70, this creates a very asymmetric risk-reward in a market where we see very few.

I. Background:

CLS is dual listed in the US and Canada and reports its results in USD.  Celestica provides Electronics Manufacturing Services (EMS), including procurement of bill of materials, PCB assembly, test, product assembly, and fulfillment.  They also provide some higher value services, like joint engineering and product development (JDM services).  Celestica was acquired by the PE firm Onex in 1996 out of IBM, and publicly listed in 1998.  Onex remains on the board and continues to own 20% of the company 20 years later (an outlier in their portfolio and any private equity portfolio for that matter).  They also continue to control the company via super-voting shares.

Celestica’s stock price recently declined due to guidance modestly coming in below expectations the last couple quarters, following a series of 4 quarters in a row of revenue growth, profit growth, and giving guidance above expectations.  EMS businesses are subject to the performance of their end markets and customers – there is little visibility so the business is very difficult to predict more than a quarter out, for both investors and the company.  This creates frequent opportunity in EMS stocks as investors get overly positive when business is performing well (“They have finally turned the corner!”) and overly pessimistic when results are weak (“This business sucks, it’s impossible to predict, I give up”).  The past few years we have seen similar business issues at Benchmark and Jabil result in temporarily depressed valuations.   They quickly re-rated as business with certain customers improved and as other company specific catalysts played out.  We expect a similar thing to happen over the next 6-18 months at Celestica.

II. Basic Primer on EMS market:

These businesses are generically considered pretty terrible businesses for the following reasons:

1)      Very thin operating margins

2)      Intensively competitive in certain segments of the market (especially consumer electronics)

3)      Consistent pricing pressure from customers as electronic components tend to deflate over time, requiring companies to constantly grow volumes to grow revenues

4)      Frequent non-recurring restructuring charges as programs churn off or companies optimize their footprint.

These attributes are particularly true of what is called “high volume, low mix business”.  A good example is consumer electronics.   Vendor selection is mostly done on price, and while quality matters, these products are relatively easy to assemble and test, carry low price points, and can be produced by hundreds of vendors.  Product cycles are also very short, in many cases only 6 months or a year, requiring an EMS vendor constantly need to win new programs just to stand still.

These attributes are less true of “high mix / low volume” business.  Examples of these industries include medical, Aerospace & Defense, and certain technology products (optical products, high end networking products).  In both A&D and Medical, the business is actually quite good: the EMS provider tends to be spec’d into the FDA / FAA approval for product production, and long product cycles (10+ years) ensure a pretty consistent flow of business.  In these industries, cost of the electronic components / PCB board are usually a fraction of the cost of the product, whereas in consumer electronics they typically make up the bulk of the cost.  Further, product reliability is very important – the device fails, and someone dies.  For these reasons EMS companies perceived to have portions of “high mix / low volume” business, especially in Medical and Aerospace & Defense, trade at a substantial premium to companies with more high volume / low mix business.  Plexus, for example, trades at 10x EBITDA, and Flex (despite having a lot of high volume biz), trades at 8.5x ebitda because they are the only company that clearly breaks out profitability of their low mix business (which tends to be much higher margin than high volume biz).  Benchmark electronics recently purchased a defense EMS / components for 12.5x ebitda. Below is the comp group of EMS companies and the portions of their business that are “high mix / low volume”.  There is some judgement required here so can follow up on this in Q&A as to how we got here.

Despite the industry’s reputation for being a bad business, we think – while not a great business – this is generally a better business than most people give credit for.  EMS companies are expert in managing very complicated supply chains, not a core competency of many of their customers, who tend to view their core competency as product design or sales and marketing. EMS companies also have substantial purchasing power advantage over many of their customers.  Switching costs are high on a per program basis, though more limited on new product launches, making industries with long product cycles particularly attractive.  Price, especially in “high reliability” products like Medical Devices & Aerospace and Defense, is not as important as reputation and quality.  EMS companies take little/no inventory risk, much like many distributors, who operate on similarly thin margins but tend to trade at substantially higher multiples.

We can debate the quality of the business, but given CLS valuation and the valuation we are underwriting (5.5x ebitda) you don’t’ need to believe much.  One thing you can’t debate, however, is the historical ROIC of Celestica and its peers, which if it was all you saw would make the industry look pretty good:

We define ROIC the way textbooks do, taking no adjustments: NOPAT / Invested capital.  For NOPAT, we do not back out restructuring charges, impairments, or stock based compensation.  For invested capital, we do not back out intangibles or goodwill.  Adjustments of this nature would yield ROICs in the high teens / low 20s.

III) We believe Celestica is a very cheap stock, any way you slice it

One way we think about this business is separating the growing, sticky, “high mix / low volume” business from the legacy, declining business, which is how FLEX has segmented its business.  For simplicity, let’s call these businesses “Good Biz” and “Bad Biz”.  If you view the business on this basis, you get the Good Biz at about 7x EBITDA of EBITDA and ~400M of WC in the $150M EBITDA Bad Biz for free. Below is our illustrative, best guess view of the business mix and breakout of profitability:

We’d argue you should at least give them credit for WC in the legacy biz (essentially putting about 2.5x EBITDA on it).  If you do that you are buying the Good biz for 4.5x EBITDA.

So what is this business you are buying for 4.5x ebitda?  We think this is a secular grower which has organically grown revs at an 11% CAGR since 2010, with high single ebitda margins and a mid-teens ROIC.  The company breaks out this business historically, at least on revenues.  We’d argue this business would probably fetch at least 10x in a sale or if it was public as a standalone company (ala Plexus):

In addition to be exceptionally cheap on absolute basis, Celestica is also incredibly cheap relative to peers, as FLEX/BHE/SANM have all benefited from investors increasingly positive perceptions of high mix / low volume business.  Celestica has yet to experience this same re-rating despite having similar business mix to peers:

IV.  Why Celestica is cheap and why we think the discount is unjustified

1. Hidden assets

We believe the company has over $225M or $1.50/share of “hidden assets that should be converted to cash within the next 12-18 months that the market doesn’t give the company credit for.  These include:

A)      Corporate HQ ($95.6M) – Celestica’s headquarters is a large facility located in an attractive part of Toronto.  In addition to corporate offices, they also have manufacturing operations in this complex.  In July 2015, the company agreed to sell the property to be re-developed into a mixed use property.  They received a down payment of 10% and expect to receive the bulk of the balance once re-zoning is approved, which is expected to happen in the first half of 2018. See the company’s 10K for more detail, or the below article:  https://www.bisnow.com/toronto/news/mixed-use/celstica-reaches-deal-to-sell-hq-site-to-development-consortium-48491

Because this value is not on the balance sheet, and can only be gleaned by reading the 10K, we think investors are missing this in their valuation of the company.

B)      Normalization of inventory ($105M) – the entire electronics supply chain is currently experiencing shortages of certain components.  When this occurs, EMS companies are left with partially completed PCBs and also try to stock up on more components than they typically would do avoid production delays.  The company has 12 month “put” options with many of its customers on excess inventory, so takes minimal inventory risk.  Days Inventory in the last quarter was 65 days versus historical averages of closer to 50 day.  We view normalized as 57.5 days given increasing shift to A&D business which carries longer working capital cycles.  We believe days inventory should reduce in the next 12-18 months, though timing is uncertain and it’s possible this gets a little worse before it gets better.

C)      Solar assets held for sale ($24M) – the company had a solar division which they have shut down. They own several facilities and PP&E they are in the process of selling. We discount the balance sheet value here by 15%.

2) Business mix better than it appears in segments normally considered “high volume / low mix”

Based on multiple calls with industry experts, we believe the bulk of Celestica’s business is actually “high mix / low volume”, just not in the industries investors typically look for it.  This business mix difference is notable when comparing Celestica’s “traditional” EMS business to peers.

Let’s take Benchmark electronics as an example.  Telecom is 12% of BHE’s revs.  In telco, one of the main products they produce are set-top boxes.  A set top box is literally a PCB surrounded by a cheap plastic box, and is produced in units that measure in the hundreds of thousands.  ASPs are low, $100 or less.

A decent chunk of Celestica’s telco business is from Cisco.  Cisco has 4 EMS vendors – Flextronics & Foxconn do the high volume / low mix products, while Jabil & Celestica split the high mix / low volume products.  We’ve spoken to several Cisco formers who describe the relationship as very strong / sticky and Celestica’s capabilities as unique.  For a simple home electronics product, you may have a PCB with 4 layers of copper.  This makes it relatively easy to assemble and test.  For a product like their CAT-9300, which retails for over $9,000, there are often 28-36 layers, which are all susceptible to crosstalk & noise.  You have to have very precise manufacturing and test capabilities to produce these well, it’s just a different animal making these products vs. set-top boxes.  Poor quality leads to bad yields and failures in mission critical systems where your customers can’t afford to have them.  So while this is typically in a category that is considered “bad” or “high volume / low mix”, it’s actually much better business than it appears to be, just in some cases with no (or negative) secular tailwinds.

3. Lack of appreciation for structurally growing parts of the business

We have identified two main parts of Celestica’s business, considered to be shrinking / bad business which we believe are overlooked and are more fairly classified as higher value / growing.  In our valuation we are valuing these at working capital, but an argument can easily be made to put a real multiple on them.  We also think their diversified segment (which mostly consists of their aerospace and defense business) is better than people give it credit for.

A)   White box business (~5-8% of revs, growing 20%+)

Most technology hardware is increasingly becoming commoditized.  As many areas of the hardware establish standards and de-bundle software from hardware, there is increasingly less value added from buying branded hardware products, especially in servers & networking.  This has led to the rise of “white-box” products that are joint developed between large cloud companies (Facebook, Amazon, Google, etc.) and manufacturers. These companies are increasingly bypassing large OEMs like Cisco and going direct to JDMs / ODMs / white-box providers.  In other words, whereas in the past the value tended to accrue to the OEM, increasingly the OEM is being cut out and the EMS can go direct to certain large customers and capture some additional value for themselves.  Although less than corporate margin today as the business ramps, we believe Celestica’s white-box switch, server, and networking businesses are amongst the largest in the industry and growing rapidly.  This business should serve as a natural offset to secular pressures in certain product areas of theirs, and over time should be margin accretive.  We believe this business for Celestica today is around $300-500M and growing double digits.  Celestica, for example, has nearly 30% of the white-box switch market per IDC, which forecasts growth of 35% in 2017 in this market.

It’s also worth noting that Celestica has some very powerful partnerships in this area of their business with meaningful players that the company has not widely advertised and that investors don’t appreciate.  Celestica, for example, is Facebook’s engineering & manufacturing partner for their white-box optical product, Voyager:

https://code.facebook.com/posts/1977308282496021/an-open-approach-for-switching-routing-and-transport/

B)   Optical business (10-15% of revs, growing double digits long-term)

Celestica is one of the leading EMS providers of Optical Systems.  Their competitors in this market include Sanmina and Fabrinet.   We estimate somewhere between 10-15% of Celestica’s revenues come from Optical. Optical has several unique challenges that make it very difficult to assemble and test relative to other types of products.  It is also an end market that is growing rapidly driven by greater demand for higher bandwidth speeds from telco’s and datacenters.  Although the optical market will be weak this year, we forecast the market to grow about 15%/yr for the foreseeable future. By contrast, Fabrinet has grown its revenues at a 23% CAGR since 2012 and has done so with the highest ROIC in the EMS space.  Optical revenues will be off quite a bit over the next couple quarters, but is forecast to return to growth in the next 2-3 quarters, which is part of what is creating the opportunity in Celestica today.  We’d note that this is not an issue specific to Celestica – Sanmina and Fabrinet have also noted similar weakness, yet trade at substantially higher multiples.  After growing nicely the past few quarters, Celestica’s telco business is expected to be down in Q4.  We believe that much of the “beats” and recent “misses” in Celestica’s business stem from the strong growth and subsequent weakness in the optical market.

C)   Aerospace & Defense Business (20%+ of revs)

Celestica’s main piece of traditionally accepted “high mix – low volume”, or “good business”, is their Aerospace & Defense (A&D) business.  Here, Celestica makes electronic components and subassemblies that go into Aerospace and Defense (A&D) applications.   This business is attractive for several reasons:

1)      A specific manufacturing facility tends to be specified into regulatory approval for many A&D products.  As you can imagine much of what they are involved in his highly regulated, so switching costs are exceptionally high / it doesn’t really ever happen.

2)      As a portion of the product cost, the PCB and electronic components tend to be minor costs and a minor part of the process in A&D.  Honeywell, whom we believe is in the process of outsourcing most of their EMS work to Celestica, is a good example of how Celestica is able to add value here.  Honeywell does $40B of revenues a year.  Based on conversations with formers of Honeywell, we believe their cogs associated with electronic components is about $500M.  By comparison, Celestica spends over $5B a year on components and has expertise procuring them, leading to substantial procurement savings.  On top of that, because all Celestica does is assemble electrical components, it’s fair to say they are better at this than a business like Honeywell where this is a very small part of what they do and a core competency.  We believe Celestica can deliver 20% savings to Honeywell while also earning a high single digit EBITDA margin and a high teens to low 20s ROIC.

3)      As we understand it, Honeywell is the only Aerospace and Defense company who has outsourced a material portion of their electronics assembly to an EMS company.  Celestica won the majority of this business over several other notable EMS vendors in a bakeoff about 3 years ago, including from Plexus, which was an incumbent provider.  We believe Honeywell has a plan to transition nearly all their EMS business to Celestica over a number of years, with a total value in excess of $1B.  We still think they have substantial embedded growth from this contract even though it was announced a couple years ago, as business takes time to transition.  We think A&D is still in the very early innings of outsourcing EMS work and that Celestica is in prime position to do very well in this market given their Honeywell win.  Further, we would note Celestica’s CEO used to be head of procurement at Honeywell and has a very strong A&D background as well.

4. Excellent financial discipline, ROIC focused culture

Based on our conversations with former employees, the consistent criticism we have heard of the company is that they are “too conservative” and “too ROIC” focused.  Onex and the leadership have instilled this culture deep within the company.  Former sales people we spoke to complained about how buttoned up Celestica’s RFP process was to ensure that an appropriate ROIC would be achieved and how this caused them to lose out on certain business.  Other EMS companies we have heard will go after business with a less clear sense of return on capital, because they want the biz or believe it to be “good” biz without a clear sense of ROIC on the business.  Plexus, for example, despite having arguably the best business mix in the industry, also has consistently had one of the worst ROICs in the industry.  Nearly every former employee we spoke to at Celestica proactively mentioned ROIC, which we found unusual.  ROIC is displayed prominently next to EPS in every earnings press release.  ROIC determines roughly half of management’s incentive compensation, with the other half being driven by total shareholder return.

d) Limited growth historically but excellent history of FCF

The historical financials for Celestica and most EMS players are not confidence inspiring from a topline standpoint – revenues and EBITDA are basically unchanged since 2009 in Celestica’s case.  That said, this is not unique to Celestica (e.g. see BHE or FLEX).  This isn’t all bad, either - mix has changed in a positive way.  In 2011 consumer was 25% of CLS revenues and Blackberry was the company’s largest customer.  Today consumer is close to zero.  That business has been replaced with increasingly large chunks of higher value / higher margin business.

We’d also note that over the last 10 years, the worst drawdowns the company had on an EBITDA basis were 15%, in 2009 and in 2013.  Historical cash flow production is also excellent for a business trading at this valuation:

V.  Key Risks:

Timing: It is very possible we are early and that results get worse before they get better.  Our insight into each line of business in terms of understanding why it is currently weak and when it will bounce back is imprecise at best.  At the same time, our experience in this space is when companies have hiccups or show declining revenues, everyone gets pessimistic about their long term prospects, and when they are doing well and growing, people tend to project that growth forward in a straight line that won’t come to pass.  We generally have done well buying these names when results aren’t great, and selling at some point in the future when results are better, but trying to predict the timing is near impossible.

Acquisition:  The CEO is public about wanting to do an acquisition, potentially a large one, and potentially at a healthy multiple.  Based on our conversations with the CEO and our understanding of the board, we think it is highly unlikely they do a “bad deal”.  Also, because some investors value the company on a P/E basis, capital deployment would almost certainly be accretive on an EPS basis and would likely be considered a positive. We’d also expect an acquisition to improve the “good business” mix and that investors would be more likely to award the company with a higher multiple in that case.  That said, we’d rather they do more simple things like give disclosure on EBIT mix, as FLEX has done, which we believe would show a business with the majority of EBIT coming from traditional “high mix / low volume markets”

Customer Concentration:

Cisco is 19% of revenues.  Although we believe the relationship there is strong, the switching costs high, and material declines unlikely, it’s always possible.  We’d note that we believe this business is below average company margin and in Cisco’s higher margin / healthier products.

Catalyst

**Better disclosure allowing for sum of parts

**Continued cash flow & stock buybacks

**Continued capital deployment at a solid ROIC in A&D growth opportunities

**Potential ONEX exit

**Parts of the business bounce back and people once again believe the company has turned a corner on growth