Such is the life of a short-seller. The biggest short in our fund (First Solar) came out last night (Australian time) and – again to use the Australian vernacular – hit us around the head with a bit of 4 by 2.
It was ugly. The stock was up almost 18 percent and was the best performed stock in the S&P. Moreover it has been up for a few days prior to the result. This has not been a profitable position.
On the plus side we run a portfolio – and somewhat unbelievably the customers performed about with market despite this blow. Obviously Bank of America – our biggest long – was up 5 percent – and that position is more than double the size of our First Solar short. Also – and less obviously – our next biggest short (which I will refrain from naming) was down hard. Those two more than offset First Solar and the rest of the portfolio was merely OK. [Even the Spanish bank short – also a largish position – did not hurt us despite Santander’s blowout results.]
If you just looked at the aggregate you would not even notice the “First Solar Pain” – but I don’t look at the aggregate – and I am hurting – if only via wounded pride. So it is time to have a re-examination – a look at what went wrong or did not go wrong. After all I need to know whether to persist in this loss-making position. As I introduced you (dear reader) to my initial thoughts I will introduce you to my continuing thoughts.
When to persist with a loss making short
There are three reasons for covering a short – and only one of them is a happy reason.
The first reason is that the thesis has played out and you have made your profit.
The second reason is that your thesis is wrong. At Bronte Capital we have a strict “no broken thesis” rule.*
The third reason for covering a short – one that happens all-too-often – is that the short poses too much risk and must be reduced for risk control. This happens with shorts but not longs because when a long goes against you (that is down) it gets smaller and hence does not threaten the fund. When a short goes against you (up) it gets larger and hence often must be reduced for risk control reasons. This is the main practical difference between shorting and traditional long investing. Positions on which you are wrong shorting hurt a lot because they get larger and you wind up wrong on larger positions. (Leveraged long investing – which we do not do – requires risk management similar to shorting.)
Anyway – there are only two reasons I will cover the First Solar short after this beating. One is risk control – and that is more-or-less automatic. If the position is too large we will trim it. The other is that our thesis is wrong.
I am not going to tell you about the risk control process (unless you want to become a client). I will tell you about the thesis.
Our thesis – and the the First Solar results
Part A - is that crystal silicon modules (c-Si) are becoming cheaper to manufacture because the Chinese are getting good at producing them and that the most expensive ingredient (polysilicon ingot) is becoming substantially cheaper as the market has become competitive and glutted.
Part B – is that the Chinese c-Si producers are competitive and that over time the price of modules will come – through normal competitive pressure – down to near the cost of c-Si cells – that is costs will determine prices.
Part C – is that the lower price of c-Si modules will compress First Solar’s margins (First Solar uses a thin-film CdTe technology) and competes almost entirely with c-Si manufacturers. Moreover First Solar will not be able to reduce its costs because it is so efficient already – essentially that FSLR’s “roadmap” to lower costs is nonsense.
Bluntly – it is Part B above that I am wrong on. The price of modules received by FSLR was very good and margins remained fat (and indeed got fatter). In the original thesis Part B was the bit without a time-frame – and I might be right or might be wrong on it in the long run – but I am most certainly wrong on the basis of the last quarter.
Lets deal with the bits I am right on so far -
Part A: There is no doubt that c-Si cells are becoming cheaper to produce. First Solar said “the low end of guidance has resilience to $40/kg poly and c-Si processing cost of 75c/watt by Q4 for non captive demand”. (See p23 of the results presentation.)
I am going to come back to that sentence because I think it is central to what has happened – but for the moment note that these are assumptions about competitor costs that were unthinkable only 18 months ago.
The cheapest of the competitors (say YingLi) will be below 75c/watt by Q4 – but it is unlikely the average Chinese manufacturer can get to that target. As for the poly price I think it will be below $40/kg – but it is not there yet. That said – it is obvious that they need to deal with competitors with much lower cost structures than previously. The first part of my thesis is not broken.
Part C: The company has an aggressive plan (“the roadmap”) for reducing their costs per watt by the end of 2014. This plan was illustrated in the following (not to scale) picture:
Note – the vast bulk of the cost-saving (18-25 percent) was to come from “efficiency”. Well the conversion efficiency of the cells remained unchanged at 11.1 percent (see p.28 of the results presentation). They need – as my last post explained – to improve the conversion efficiency by approximately 15bps per quarter over four years to meet their target. They are behind schedule.
The company states its cost per watt each quarter. Usually they cite “core costs” (that is not including “stock compensation” and “ramp up penalty”). Those costs were stable at 80c/watt. This quarter they cited “total costs” in the conference call as those dropped 3c/watt to 81c/watt.
They can’t meet their targets by dropping non-core costs to zero – they have to improve their core costs – and the main way that they plan to do that is via increasing conversion efficiency – and they are not on target. The third part of my thesis is thus not broken.
The problem part of my thesis
The problem part of my thesis is pricing. And boy is that broken.
In the fourth quarter of 2009 – a quarter that disappointed – First Solar had operating profit of roughly 145 million and production of 311 MW. Some of the profit is for their development business – but – at a first approximation it is reasonable to think of this as operating profit of 47c/watt.
In the first quarter they had operating profit of 191 million and (nicely increased) production of 322 MW. This is – at a first cut – operating profit of 59c/watt.
Now operating costs per watt only changed 3c/watt (and all of that was “non core”). Operating profit changed 12c/watt. The company shot-the-lights out on price received.
Unless there is something peculiar about First Solar this will probably apply to every solar producer – including the marginal Chinese. But whatever – competition has not driven down prices at least yet.
Competition and prices
It is a core part of my world view that competition tends to squeeze margins – but it has not happened here. There are short-term reasons that are easily identified. The most obvious is that Germany has a concessionary feed-in-tariff regime which will be adjusted (down) from the end of June. There is massive demand in Germany for panels to be installed by the end of the second quarter. First Solar is sold out.
That sort of demand pressure (sudden, urgent) is enough to delay what (I hope) is the inevitable reduction of prices to reflect lower Chinese costs.
However whilst it is part of my world-view that competition tends to squeeze margins it is not universal. Some things just don’t have much competition – I never notice margin squeeze at Microsoft. Also some things that are ostensibly competitive (beverages) seem to maintain fat margins for very long times (for example Coca Cola). I don’t see any particular reason why the third part of my thesis won’t be right eventually but I am open to persuasion otherwise.
The prices received
I said I would get back to the mathematics of the guidance – in particular the comment that: the low end of guidance has resilience to $40/kg poly and c-Si processing cost of 75c/watt by Q4 for non captive demand.
Now c-Si cells use about 6.5 grams of poly per watt (a number that is reducing). This means that the guidance is resilient until the c-Si makers have costs of (75c plus 0.0065*40c)/watt = $1.01 per watt.
Now c-Si cells should (and do) sell for more per watt than other cells (such as amorphous silicon) because c-Si cells are cheaper to install due their higher efficiency. I thought that difference should be about 15c/watt – but I have seen numbers as low as 9c/watt and as high as 30c/watt. The usually accepted number is about 25c/watt – but – to make the case as favorable to FSLR as possible I will chose a low (10c/watt) penalty.
If we get to prices being set by C-Si costs at $1.01/watt (plus say 8c/watt profit for the Chinese producers) then we should get to 99c/watt (including profit) for FSLR. FSLR’s costs are at 80c (core costs) and for the moment look pinned. This implies margin at FSLR at 18c. This could be higher if FSLR reduces costs – but is unlikely to be dramatically higher.
FSLR’s margin in the first quarter was 59c/watt. Bluntly – if the Chinese get to the costs on which the FSLR’s guidance is based then FSLR’s profit crashes if FSLR has to match the Chinese prices.
It would be hard to see how FSLR can maintain its guidance under these conditions. Except that FSLR gives us an out – they say that they are resilient “for the non-captive demand”. FSLR can maintain higher selling prices because it has “captive demand”.
FSLR has captive demand because it is also a project developer and it purchased another project developer (Next Light Renewable Power) very recently.
The question is – is it sustainable to be selling to internal or captive power developers at prices substantially above Chinese costs? Obviously it is in the short term because the solar panel market is “sold out”. It is not so obvious in the long term.
In the long term either (a) the prices at which First Solar sells to “captive demand” has to match the Chinese prices or (b) First Solar is investing its balance sheet in high cost (and hence uncompetitive) captive projects.
Obviously this can continue for a while – whilst prices remain high for solar panels. But I think it will end – and FSLR’s earnings will crash when it ends. I think Part B of my thesis is only temporarily wrong. So I will (subject to risk management) keep my position.
The market of course has a different view – and that difference is painful. Having a short go up almost 20 percent on a result is – to put it mildly – unpleasant.
*The no-broken thesis rule was detailed in David Einhorn’s book on Allied Capital. Its funny but in a book about a more-than-average difficult short the thing I most benefited on was some (very) wise words on portfolio management.