At 5:00 AM on November 12th, Canadian Solar reported the best earnings in its recent history. The company earned $1.75 per share on expectations of $1.16 – or 50% better. The revenue was $914M versus guidance of $810M – again, much better.
Pre-market trades did not indicate trouble; instead, they were showing appreciation for huge milestones achieved by the company. The stock, which closed at $31.47 the day before, traded as high as $34 in pre-market and opened as high as $33 during the regular session. Then, suddenly under unstoppable pressure, during a single day, it traded over 21M shares and dropped down 10.77% to close at $28.08. The next day, 11M shares traded, and the stock closed at $25.73; another 8.3% of the company’s value was lost. In the aftermath of the best result, the company has lost 19% of the value, or $5.74 per share. The market capitalization lost $344M.
What caused this? I looked at the headlines of the day to understand how business news described the situation. Investor’s Business Daily headlined “Canadian Solar Guidance Disappoints And Stock Tumbles,” calling the company’s fourth-quarter guidance of $925 to $975M as light, versus “Analysts polled by Thomson Reuters had expected $977.4 million.” The $2M off the guidance seemed too thin of a reason for the $344M loss in value.
Barron’s Asia published “Why Axiom Capital Thinks Canadian Solar Is the Best Solar Short,” enlisting Axiom Capital analyst Gordon Johnson to share his view: (caps are Barron’s) “Our analysis suggests the company will likely earn LESS than $2.50/shr (2015 my note).” “Why? Well, as shown in Figure 1 below, CSIQ earned $491M of project (Total Solutions + EPC) revenue in C3Q14, which leaves them with just 206.5MW of on-balance-sheet Total Solutions projects (i.e., non-EPC legacy profitable projects). They are guiding to C4Q14 revenue of $925-$975M, and C4Q14 shipments of 810-860MW (at an 18% gross margin). However, some of these MWs will represent additional legacy Total Solutions projects sold in order to arrive at the company’s C4Q14 aggregated revenue guidance of $950M. We assume ~173MW, or the lion’s share of what’s left in the Total Solutions backlog (i.e., 83.8%). More so “Keep in mind that Total Solutions are the Canadian projects CSIQ purchased years ago when natural gas prices were MUCH higher. Thus, the margins on these projects are ≥25%. However, the margins on CSIQ’s EPC business are, at best, ~10%; and, the margins on module sales are MUCH lower (see CSIQ’s earnings in 2012/13 before they were selling these legacy projects). Thus, given CSIQ, by its OWN admission, as of today, has just 206MW of high-margin total solutions MWs left to sell, yet, given its C4Q14 guide, will sell ~173MWs of these ‘golden tickets’ in C4Q14 in order to hit their guidance, the company is about to see its margins COMPLETELY collapse in 2015, as well as it rev, EPS, and cash flow.” Conclusion: “Stated differently, we ask someone to show us where CSIQ is going to find the projects in 2015 (as is currently priced into Street exists) to replace the ~$1.3bn/465MW in Total Solutions sales made in 2014 at ~25% gross margins. The answer? They will not, and their earnings are about to collapse.”
Ok, that sounded awful if it was the case. Canadian has third-party business, which produces at least 15% GM, as do most of its peers. It also has a pipeline in Japan. However, Mr. Johnson had an opinion on Japan and Barron’s provided it: (caps by Barron’s) “JAPAN WILL SAVE THE DAY RIGHT?: In short, NO. While there are many questions around what the margins will be in Japan (we need to find out what the new solar policy in Japan will be; if you’d listened to what the Japanese gov’t has said up until now, there are going to be very draconian adjustments lower), let’s give CSIQ credit for the full 79.4W it is expected to sell in Japan in 2015 at its targeted 25% GM (we believe this margin is WAY TOO HIGH, but do this for illustrative purposes). Assuming an ASP of $3.00/watt, that’s $238mn in rev and $59.6mn in GM – FAR short of the $1.3bn in rev and $342.4mn in GM Total Solutions will likely contribute in 2014. Thus, assuming ALL of the 312.6MW in Canadian EPC backlog projects are sold in 2015 (at a ~$1.25/watt ASP and 10% margin), that’s another $390.8mn in rev, and $39.0mn in GM. Lastly, assuming all of the 48.8MW in U.S. solar projects are sold at a very aggressive U.S. project ASP of $2.50/watt (and a 13.5% GM), that’s another $122mn in revenue and $16.47mn in GM.”
I have reviewed information from Canadian before and after the news release, and at first look, the data provided by Mr. Johnson seemed off. Yet, presented with such a level of detail, the first emotion overcoming me was fear – I have missed this. Clearly selling was highlighting that Mr. Johnson has discovered this and those who were selling knew it beforehand, and others as they learn it, had no choice but to sell.
Overwhelmed, I wanted to see more information, but found only conflicting perspectives.
Goldman Sachs analyst Franke He wrote: “Despite an upward revision to module shipment guidance from 2.5-2.7GW to 2.73-2.78GW for 2014, CSIQ guided for its gross margin in 4Q14 to be 17-19%, vs. 22.9% in 3Q14, mainly due to anti-dumping and countervailing duties imposed on US modules and currency depreciation in Japan and Europe.”
Frank He revised the Q4 target to $1 EPS and increased EPS for 2016 by 21%. The price target ended up being $30. This would produce $4.02 in 2014 and 7.4PE. I could not get how Q4 GM was a problem if Q3 GM was up 1.9%, and how come EPS estimates were increased on collapsing margins? It appears that Canadian Solar investment evaluation had a three months’ timeline at Goldman Sachs.
Another analyst, Northland Capital’s Collin Rusch, moved the price target to $48.00, calling fear of peak earnings unfounded. He moved Q4 EPS to $1.39, FY 2014 EPS to $4.20, and FY 2015 EPS to $4.83.
Phillip Shen from Roth Capital affirmed buy with a price target of $40, following Q3 results and guidance. He said “Given its capacity expansion plans, best-in-class brand, strong B/S, and rich downstream set of projects, we continue to believe that CSIQ is one of the best companies among its peers.”
How can one company have such opposing views? More reports pointed to the upside.
Canaccord’s Josh Baribeau believed Q3 results were solid, but explained that investors were disappointed when the company did not provide details of a Yieldco setup. Canaccord saw a growing pipeline and continued significant volume for Canadian Solar. The price was lowered to $46 from $48.
Paul Coster from JPMorgan lowered its target to $36, expecting lower margins and depletion of Canadian projects. He described de-risked $3 per share for 2015 delivered from those assets and about $3.92 for 2015, overall.
In summary, nothing mentioned the concerns from Axiom; all other analysts basically suggested buying the stock. In reality, all the selling was produced by Axiom’s hypothesis about 2015. This alone was odd. Despite five analysts having an average price of $40, the market did not care; it followed Gordon Johnson‘s view of the company. It was more disturbing that the sentiment-winning thesis used by Mr. Johnson was built on a set of errors and stretched assumptions. I concluded this by doing my own analysis, but I also learned that Nomura had blown it to pieces.
In my analysis, a look at the Q2 2014 results showed the Canada-based project asset list to have 315.2MW of its own projects and 219.6MW of EPC projects. Mr. Johnson observed that there are 206.5MW in its own projects and 180.4MW of EPC projects at the end of Q3 2014. Therefore, 75MW of its own projects and 81MW of EPC projects were sold in Q3. Since the company reported 173MW of total solution sales in Q3, around 17MW the US plant sales were also recorded. The first spurious assumption by Mr. Johnson placed the lion’s share of Q4 sales to be 173MW of its own projects, referring to it as 83% of 206MW. This was certainly not a model of Q3 sales, as shown above. I have no idea how Mr. Johnson came up with it, as the company indicated five projects from Canada to be sold in Q4, which in a simple mathematical review of all projects shows around 70MW. This leaves the company with around 130MW of its own projects in Canada for 2015. So, not having any projects was a false statement.
I also checked how many of its own projects Canadian sold in the first three quarters of 2014. In Q4 2013, the data sheet for Canadian projects had 338.7MW in its own projects and 138.6MW in EPC projects. Since there are 206.5MW of its own projects left at the end of Q3, this means that in three quarters, Canadian sold 132.5MW of its own projects.
From Q4 2013, the company added the Samsung II project worth 140MW EPC, and adjusted up the Samsung I project to 133.6MW from 128MW EPC. The starting point for EPC at the end of Q4 2013 was 138.6MW. After adding 5.6MW and 140MW, the Q3 roster ended up with 180.4MW EPC. This means 103.8MW of EPC projects were sold in 2014. Canadian does not have 312MW of EPC projects left for 2015, as Mr. Johnson has declared; it has 180.4MW left at the end of Q3. This was getting frustrating, as another “error” was made to show lower gross margin.
I reviewed each quarterly news release to add Total Solution MWs, and the number came up to be 292MW. The breakdown of the Total Solution sales for 2014 was 132.5MW of its own projects, around 103.8MW of EPC, and 55.7MW in the US. It is important to remember that the exact breakdown gave the company $3.02 EPS for three-quarters of 2014, including third-party module sales.
I used guidance for Q4 to deduct 70MW for its own projects in Canada, I assumed 40MW in EPC, and 28.5MW in the US (announced today) will be sold in Q4. Therefore, the numbers in the Canadian portfolio left for 2015 are 130MW in its own projects, 142MW in EPC, and around 76MW in the US. This is clearly more than sales in three-quarters of 2014, which contributed to $3.02 per share. Why then does Mr. Johnson suggest $2.50 for 2015? Why does Mr. Johnson see only 48MW in the US if the backlog would have 76MW, as per the company’s presentation? The slightly higher volume in EPC and the US projects could cause the fluctuation in the 9 months of 2014’s 19.7% GM, but how could a 1% drop possibly be considered a collapse and compared to gross margins of 2012?
Since the company is expanding its capacity by 20%, I thought this would add revenue on scale and would drop overall margins, but it would still add net income per share. So, perhaps the missing 70MW of projects from Q4, replaced only by module sales, would drop 2015 margins? This assumption would ignore the 200MW of new projects mentioned by the company, which I viewed at least at 17% gross margin. $2.50 per watt sold at 17% GM produces $0.425 per watt versus profit of $1.25 in Canada at $5 sale and 25% margin. So, Canadian would have to sell 2.9MW elsewhere to replace 1MW in Canada, for a total of 203MW, exactly what CSIQ is planning to do. This does not even mention 25% GM for 80MW of Japanese projects planned for 2015.
I have not mentioned Japan, as I believe Japan’s pipeline will form the base of securitization, like yieldco. Michael Potter, CSIQ Senior Vice President and CFO, confirmed this view during the third-quarter conference call, but also talked about margins in Japan: “Just as a general reminder when people ask about if we’re continuing the build and sell model in Japan, we’ve said in the past that the gross margins for the Japanese projects are at least as good as the Canadian projects and the ASPs have been increasing over the last year. So, we certainly think that there would be a good source of revenue and gross margin if we did not do a securitization. That is not our first choice, but would support our current model if we chose to do that as well.”
Nitin Kumar from Nomura established that ASPs of projects sold to date in Japan have an average of US$4 per watt, a dollar more than Mr. Johnson’s and at a construction cost of US$3 per watt, 25% gross margin. The difference is $0.25 per watt to the bottom line from Axiom, and in the case of Canadian this could be even higher due to its own supply of modules. Why does the market think that Potter and Kumar are wrong, and Mr. Johnson is right about doubt on Japanese margins?
Mr. Johnson also refers to METI changing FiT. Canadian has already posted its FiT rates for all projects acquired and METI review will produce adjustment in early 2015 to new projects. Those METI concerns created by Mr. Johnson Nomura calls “talking about draconian law changes which are something the METI has so far not talked about.” Nomura added: “Yes they are reviewing the FIT policy but there is no indication that the changes are retroactive. Also, the bottleneck for installers to construct projects is not their lethargy, but the fact that Japan solar utilities have been slow to give connection permits to large projects. So, to create a fear-factor on something that Japan authorities have not hinted is patting your own back.”
While Canadian stock is selling off, Brazilian assets of 114MW could become part of the same securitization portfolio as the Japanese ones. On November 28th, Solatio, Canadian’s partner in 114MW of projects, will be bidding its 270MW in another auction. I do not know if Canadian is involved, but I speculate it can be, judging by their interest in the country. Mr. Johnson did not cover this.
Due to smaller Total Solution contribution, and also currency fluctuation, gross margin will be unfavorable in the fourth quarter of 2014, agreed. One thought comes to mind that makes those margins lower, and I am surprised that analysts do not see this possibility. I believe that Canadian will be selling to China in Q4, probably in a large percentage and as much as 50% of the 720MW destined for third-party module sales. The logic behind it is simple: ASP in China is expected to be the best in the fourth quarter out of the entire year. Another observation is that selling in China can trigger deferred tax asset line and value-added-tax recoverable, in the same fashion as JinkoSolar had done. A portion of the $88M in assets on Canadian’s balance sheet can be used against earnings in China, a domicile of subsidiaries with the largest carry-forward operational losses, an item that is the biggest dollar contributor to DTA. This can, if exercised, produce a lot more to the net income then the average estimates depict.
In summary, I concluded the only fear factor to sell the stock was built on errors and forged assumptions. In the absence of good reason, someone made sure to go for volume. 32M shares had a jaw-dropping effect on fear-stricken investors. Moreover, the market behavior instilled doubt about the power of earnings and added anxiety about how irrational this behavior can be in the future. This is all with some purpose unknown to me, but with a known outcome of lower stock price.
So, how do I move forward? At the dawn of the new day, Canadian Solar is a very good company, and has not stopped being one just because of its share price. Factoring for fear has been part of the industry from its beginning. From being solar, Chinese, FiT dependent, oil price linked, tariff restricted, real and not so real, fears coxswained the stock price. In the case of this fear factor, almost always, volatility will settle down on value. Investors must continue to focus on real details and never stop believing in power of financial statements.
