Description

LDK Solar is a potentially fraudulent, but in any case poorly positioned and over valued, producer of polysilicon wafers for use in photovoltaic (solar) cells.  As investor’s focus on the underlying economics of LDK’s core business rather than claims about above-industry margins and backward integrating into polysilicon manufacturing, the stock should experience considerable multiple contraction (currently 55x my 2008E EPS estimate) and estimate revisions.

Production of wafers is essentially the second step in the solar value chain. The first step is the production of polysilicon from sand or quartz; this polysilicon is then used to make wafers; wafers are used to make solar cells; solar cells are then put into modules that are ultimately distributed and installed for end use.  The entire solar industry is currently constrained by a lack of polysilicon supply. Polysilicon has two uses – it can be turned into wafers for use in either semiconductors or solar cells.  As with many commodity industries, polysilicon capacity was overbuilt in the 1980s as a result of the explosion in semiconductor demand.  The influx of new demand from solar has resulted in a supply/demand imbalance that has driven the spot price for polysilicon from under $50/kilogram a few years ago to $300+ (depending on the grade) today. While a good deal of polysilicon capacity is set to come on line next year, industry participants do not anticipate any slackening in spot prices for the raw material in the face of large and rapidly growing demand for solar cells and modules. 

LDK is a compelling short because one does not have to take a view on overall long term feasibility/sustainability of solar demand to realize that the company is in a very poor strategic position that will lead to eroding economics over time. Personally, I am highly suspect of any industry that requires heavy government subsidies to stimulate demand. Neo628 recently posted an writeup of FSLR that nicely summarizes the economics of the solar industry and the potential problems it may face going forward that I’d recommend to anyone interested in the space.  If one has a negative view on solar demand, pretty much everything in the value chain will make a good short, from high multiple polysilicon producers such as WFR to solar module manufacturers such as FSLR. But LDK’s value proposition leaves it poorly positioned even in a world where solar demand thrives. 

To reach “grid parity,” the installed cost of solar needs to drop by ~50%.  Module, wafer and cell manufacturers are in a race to add scale in order to lower costs.  As a result, wafer prices are under constant pressure; LDK management has guided investors to expect quarterly price compression of 1 to 2% for the foreseeable future.  In the face of this competitive dynamic, LDK is adding wafer capacity and expects to go from 400MW of capacity at year end 2007 to 800MW at YE 2009 and 1,600MW at YE 2010. This expansion will required ~$1B in capital.  Since coming public in summer 2007, LDK has put forth a steady stream of press releases announcing forward sales contracts for their wafer production.  While details on the seven contracts has announced since August 2007 are scarce, my rough calculation is that LDK has sold forward ~80% of 2008 production and nearly all of 2009 production. From a practical point of view, LDK needs to sell its forward production to finance its growth plans and this is where it runs into a big problem: LDK is locking in declining wafer prices without a locked in supply of polysilicon – they are essentially short polysilicon.

When LDK went public in spring 2007, management touted the company’s ability to utilize recycled/scrap polysilicon in its wafer manufacturing process – an advantage (in both sourcing and cost) over other players in the industry who are dependent upon raw/virgin polysilicon that led to LDK’s 38%+ gross margins.  In the two quarters since the IPO, this claim has been put into serious doubt. Gross margins declined to 35.2% in Q2 2007 and 30.8% in Q3 2007.  On their most recent conference call, management guided to gross margins of 25% to 30% in 2008.  Management has done a poor job of explaining this margin compression, but my conversations with industry participants indicate that scrap supply has tightened and prices have risen considerably over the course of 2007.  Other publications and websites (notably Barron’s) have chronicled the whistle blower investigation into LDK’s inventories (recently declared a non-issue by the audit committee) and I’ll not revisit those at length here because frankly there is not much I can add.  In my experience I am better off relying on the numbers than conspiracy theories as to what may be going on with the inventories half way around the world – and the numbers are starting to look awfully shaky.   

In an attempt to revive margins and satisfy their large polysilicon needs, LDK has announced plans to build a large polysilicon manufacturing facility.  In August 2007, the management began touting their ability to produce 6,000 metric tons of polysilicon by YE2008 and 15,000 metric tons by YE 2009, at an estimated capital cost of $1.2B to $1.8B.  This announcement set off a new round of speculation in shares of LDK as some sell side (UBS in particular) and (apparently) numerous buy side analysts began discounting massive earnings power based on LDK’s proposed new facility.  Based on my observations of the industry and discussions with numerous participants, the timeline and cost structure outlined by LDK management are essentially impossible to achieve.  Industry leaders WFR and Hemlock (private company that is a JV including Dow Chemical and Mitsubishi) both have new facilities coming on line in early 2008 – each of these entered the planning stages nearly three years ago and have taken more than two years to build.  While LDK may derive some benefit from inexpensive labor and a loose regulatory environment, the primary binding constraints to building a new polysilicon facility are capital equipment, expertise, and access to expensive and difficult to access chemicals – none of which are obviated by LDK’s location in China.  While I do not have a definitive view as to whether LDK will ever succeed in manufacturing polysilicon, I believe it is a safe bet to doubt it (just as it is a safe bet to doubt whether IOC will ever build an LNG plant or USU’s ACP project will work economically) and I believe it is nearly a sure thing that the plant will not come on line prior to YE2009. 

I believe LDK’s financial statements will be hit with a double whammy in 2008 as the margins of their core business decline and they begin to put significant debt on the balance sheet to finance both the expansion of wafer manufacturing capabilities and the green fielding of the polysilicon factory.  As 2008 earnings come in below expectations and investors begin to realize that no growth is forthcoming in 2009, I believe the market will re-rate this stock from 55x forward year earnings to a multiple more in line with industry average of ~35x at best, and potentially much lower as a non-growing manufacturer of a commodity product with no pricing power.  Absent a slowdown in capex, LDK may also run into a financing problem in late 2008 as I do not believe lenders will be eager to lend more than 4x EBITDA to a company in LDK’s position, and a dilutive equity offering could be the solution.  Should the SEC determine that the company has improperly accounted for inventory, I would expect the stock to completely implode. 

Catalyst