At its surface, Canadian Solar's (CSIQ) third quarter earnings appears weak, but beneath its headline numbers the company continued to improve on its strategic positioning within the solar industry. For investors, it is more important to determine which companies will emerge stronger from the industry's current destructive consolidation period than quantifying near term negative earnings cycles. Much like other US listed Chinese solar direct peers such as Trina Solar (TSL), Yingli Green Energy (YGE), and Suntech Power (STP), Canadian Solar continues to advance itself at the expense of smaller or less cost competitive rivals being forced out of business. However unlike its larger rivals, CSIQ has arguably executed through the current down cycle relatively much better as already noted in a previous summary.
For the third quarter 2011, Canadian Solar's shipments increased 24% sequentially to 355MW which was within the company's previous guidance. While CSIQ's quarterly shipment strength was in part due to delayed revenue recognition as explained in a previous summary, the company nevertheless continued to post market share gains even relative to some larger peers such as Trina Solar which posted sequential shipment declines during the quarter. Year over year, CSIQ's third quarter shipments grew an impressive 78% over the 200MW shipped in Q3 2010.
Revenues in Q3 only grew slightly on a sequential basis to $499.6m from $481.8m in Q2. This was much lower than my $547m estimate due to lower than anticipated module average selling prices ("asp") and lower system project revenue recognition which the company indicated may be pushed forward into early 2012. Gross margin sank to 2.4% due to an inventory and purchase commitment write down. Without these one time non-cash charges, gross margin would have been 9.6%. Although this was also lower than my 10.6% estimate, Canadian Solar's adjusted gross margin was still strong despite lower than expected module asps. Part of this gross margin strength was due to higher margins in its growing systems businesses which the company stated in its Q3 earnings conference call as being 10% higher than for its module segment. Adjusted gross margin was also buffered by much lower than expected module production costs.
On a U.S. GAAP basis, Canadian Solar posted a Q3 loss of $43.8m or -1.02 in earnings per share ("EPS"). While the two one time non-cash charges totaling $36.1m described above accounted for the majority of this loss, the company also recorded a net foreign exchange loss of $9.5m. Excluding these items, CSIQ's operational earnings would have totaled $1.8m or 0.04 in EPS. While short of my 0.16 EPS estimate which assumed higher module asps, the company's third quarter earnings was not a major surprise.
Apparently CSIQ changed its foreign exchange hedging practices in a quarter when the key currency, the euro, declined significantly relative to the usd. In the past, Canadian Solar over hedged against the euro which normally resulted in net foreign exchange gains during periods of euro weakness. Had the company kept a constant currency hedging strategy, it would have recorded a meaningful net foreign exchange gain in Q3 instead of a significant net loss of $9.5m. As a result for the first three quarters of 2011, CSIQ posted losses for both hedging as well as foreign exchange translations totaling $34.1m. Canadian Solar like many other US listed Chinese solar peers reported much higher than expected net foreign exchange losses due to an inconsistent currency hedging strategy. Companies which tend to have high foreign exchange exposure would be better off having no currency hedging strategy than an inconsistent one as most evident by Suntech's horrendous cumulative foreign exchange losses in the past two years.
Aside from near term U.S. GAAP losses witnessed by the majority of the solar industry as inventory liquidations by defunct peers continued to push asps across all major silicon based solar verticals lower, operational metrics at Canadian Solar showed surprising and significant improvements. Perhaps the most important metric is the overall processing cost for producing a solar module. This includes producing and/or procuring costs for polysilicon, silicon wafers, and/or solar cells among other raw materials used in the supply chain.
In the past as mentioned in my prior articles, CSIQ has had a very spotty record in managing production costs. More recently, the company has made significant improvements in the right direction. Whether it's due to the company's own initiatives or simply from pricing reductions across the entire value chain is unclear. Most likely both factors played a significant role in reducing CSIQ's polysilicon to solar module processing costs down to .75/watt in Q3 2011 compared to a range of .85-1.00/watt in past quarters. While processing costs at Canadian Solar still fell short of industry leaders such as Trina Solar or Yingli Green Energy which both reported mid .60s/watt figures, CSIQ targeted a .65/watt processing cost goal for the fourth quarter. In a recent January 2012 update, the company listed its overall module unit cost in Q4 at .76/watt which suggests the targeted goal had been reached. For the first time in the company's history, its real time module production costs directly rival that of industry leading direct peers such as TSL and YGE.
Despite significant real time cost improvements, lagging effects of inventory blending should continue to put near term pressure on overall margins. Since module asp declines continued to outpace blended cost reductions in the fourth quarter, gross margin at most solar companies including CSIQ should remain under pressure. Module asps were guided towards 1.00/watt, although Canadian Solar noted it would try to keep the average above this range. This is down from an estimated low 1.20s/watt in Q3. Although production costs using real time procurement was below .80/watt in Q4, the average would have to be blended with inventory costs above 1.10/watt estimated in Q3. As a result, the company guided gross margin to range between 5-8% for the fourth quarter, down from an adjusted 9.6% in the third quarter.
In terms of unit shipments, Canadian Solar guided the fourth quarter to be approximately flat on a sequential basis. The company expects shipments to range between 340-360MW vs. 355MW posted in Q3. Using this guidance as well as other metrics either stated or implied by the company's comments, a fourth quarter earnings estimate has been compiled below. Again as with all my estimates, all metrics are bound by information given by the company and do not speculate beyond these numbers. These estimates also exclude unannounced gains or charges which may result from non-operational items. Although I normally exclude non-operational items, the estimates below do include a small net foreign exchange loss as well as an inventory provision the company noted was not expected.
CSIQ Q4 Earnings Estimates:
- Revenues: $368m module + $50m system = $418m
- Shipments: 350MW module
- ASPs: 1.05/watt
- Blended Unit Costs: .95/watt
- Gross Profit: $35m module + $9m system - $8m inventory provision = $36m
- Gross Margin: $36m / $418m = 8.6%
- Operating Expenses: $42m
- Net Interest Expense: $10m
- Net Foreign Exchange Loss: $4m
- Tax: $0
- Net Loss: $20m
- Diluted Share Count: 43.2m
- EPS: -0.46
As usual, net foreign exchange estimates are extremely hard to gauge without knowing the company's exact hedging policy. The small $4m net foreign exchange loss estimated above assume Canadian Solar kept a constant currency hedging exposure with the third quarter which may not be the case. Investors should focus on operational figures which exclude any gains or losses recorded due to foreign exchange translations and only penalize companies that mismanage currency exposure if losses become chronic and large in proportion.
Clearly 2011 was a difficult year for the solar industry and Canadian Solar's fourth quarter will continue to appear negative at the headline level. Due to accounting which results in a lag period in recognizing lower blended costs during a cycle of declining asps, results for all producers will appear dire. However the downward cycle in 2011 should not be much different from prior down cycles such as late 2008/early 2009. Once prices begin to level off, gross margin among the most cost competitive companies should start to expand. As bad as solar earnings were in the solar trough post credit crisis 2008, earnings were also quick to reverse as many industry leading producers started to post record earnings in late 2009 and throughout 2010.
Although it may be too early to draw absolute conclusions, asps across most silicon based solar verticals may have already bottomed. While inventory liquidations by exiting firms caused dramatic price erosion, pricing should stabilize once these liquidations abate. Leading cost producers may be willing to lower pricing to match liquidation pricing, but they simply would not go below their own real time costs. Renesola (SOL) which was one of the first companies to accurately predict module asps could drop to as low as 1.00/watt stated directly in its Q3 earnings conference call pricing had already dropped to levels where the company would not go beyond. Research Firm IMS noted module pricing actually increased 7% in December 2011 after falling over 40% since the start of the year. While investors should not expect asps to rise in general, any sign that pricing has started to flatten would be good news since margins at the most cost competitive manufacturers should start to expand as the lag time for blending inventory costs lower become absorbed.
In Canadian Solar's case, per watt gross margins should expand above .20/watt once inventory carrying costs get blended down towards its real time module manufacturing cost of .76/watt as stated in a recent update. Assuming module asps level off and stay around 1.00/watt, overall gross margin could improve towards or even above the 20% range. The key question of course is at what pricing solar modules normalize due to demand elasticity.
Contrary to numerous views and claims generally by the Wall Street community, demand has never been an issue. Recent reports by IHS pegged 2011 solar installations grew by over 34% to almost 24GW. IMS Research was more aggressive and estimated 2011 installations grew over 50% to over 26GW while other industry research groups had numbers close to 28GW on record Germany and Italian installations. Perhaps the argument for price elasticity has been validated since demand has continued to surge as pricing declined. The issue for investors is to determine which solar companies could still operate profitably at pricing levels capable of absorbing its manufacturing capacity.
On page 15 of Canadian Solar's January update, the company notes according to firms such as Goldman Sachs and Photon, only 14GW of global capacity could manufacture at or below 1.00/watt. Although the numbers may vary slightly, other companies such as Suntech have also quoted similar metrics. Logically as long as global demand surpasses 14GW while assuming the inventory channel has been normalized, companies that can manufacture below this price should be able to sell out its capacity. Once inventory liquidations have been completely played out, module asps could even arguably rise in accordance to the production cost capacity and global demand curve. Gross margin at CSIQ along with other large scale low cost module producing direct peers such as TSL, YGE, STP, and LDK Solar (LDK) should expand perhaps not to record levels witnessed in the past but to levels where corporate profitability can once again be achieved. At recent market valuations of as low as 1-2x trailing earnings prior to the recent downturn, any potential turnaround may be completely discounted by Wall Street.