Software eats part of the world

Marc Andreessen's WSJ article about software eating the world caused debate amongst my colleagues at Bronte Capital. It's a view that informs a fair bit of Bronte's portfolio.*

You see I wouldn't want to make guitar tuners.

After all there is a guitar tuner app on my Android phone and it works well enough. Can't plug it into my son's electric guitar and tune it whilst he is jamming out some Deep Purple covers at the local pub. (Actually he isn't doing that... he is only 11 but I think he wishes he was doing that...)

But that sort of expresses the problem. For most purposes the hardware (the guitar tuner) becomes "appified" - that is it just becomes another app on your mobile phone. A software guitar tuner replaces a hardware one except for specialist uses.

And many things have become appified. The list of devices in a mobile phone is extensive (guitar tuner, satellite navigation, alarm clock, filofax, digital voice recorder, camera, my HP15c calculator etc).

But phones are just the start. Look at this:

Cyberhotel on FreePBX

OK - for the uninitiated I will explain what this is. Its a virtual computer download. After all you can run a computer on Virtual Box (or a Vmware player) and it will virtualize a computer - that is run a computer on software that looks, feels and operates like the box you purchased online from Dell. The computer you are virtualizing with this download is (and you need to breathe deeply to get the significance of this) a hotel administration system which will manage billing for telephones, wifi hotspots, the porno movie you watched at 10.30pm because you were lonely and in a strange town and the mini-bar.

And because it is a virtual machine you can house it in any nondescript linux running box provided you run Vmware on it and plug a few nondescript switches into the back of it.

The expensive hotel PABX has been virtualized: eaten by software. This "virtual machine" the product of some French company that assembled it from publicly available bits of software just wants to sell and install a few trivial phones.

Routing systems - even quite complicated ones - have been virtualized.

Which neatly describes the problem for Cisco. Cisco you see makes complicated integrated hardware-software devices - and their product set - like most hardware-software devices - is being appified.

What is happened to guitar tuners is happening to Cisco which is the real problem with the stock - the reason it trades at such low multiples.

And you can see this in their results: we are in the middle of one of the biggest routing booms you could imagine - as we go from a world where a few devices are connected to one where every device, every tablet is connected to the web. Cisco talk about 50 billion devices but you can't see it in their revenue line.

Instead in the middle of the biggest imaginable boom they had that dreadful conference call where they talked about government spending being weak. Hey isn't this a private sector routing boom? Well it is but the private sector is appifying really fast. The government sector are all paranoid about Chinese government hacking and terrorist vulnerabilities and so are wanting the tried-and-proven hardware-software integrated device (which they think is harder to hack). Government paranoia is stopping the software eating the hardware and them guys at the Department of Homeland Security are sitting there safely in their (antiquated Cisco) box...

But whilst the government holds back the tide of history the lesson holds true: if you make hardware-software integrated devices your risk is you are going to get appified and unless you do the appification someone will do it for you.

Marc Andreessen is right: software will eat the world or at least part of it.

If you are a pure hardware maker its going to be really diabolical. I should illustrate by example. A financial firm I know (one office, two floors) runs about 70 desktop computers running Windows. They used to have a computer on each desk, a series of centralized servers and a backup of the servers (minute by minute) stored off-site and the desktops backed up once a week. You were told not to store stuff on the client computer - only the server.

There were a bunch of security risks with this. For example the client computers all had USB ports so you could plug in a USB key and steal data. So the USB ports were disabled. The client computers still had hard-drives. A staff member could steal data by downloading stuff to their hard drive and then walking out with the hard-drive in their bag. I guess you could lock up the client computers and put alarms on them.

It ain't run that way anymore. The computers are now virtualized.

I need to explain that. With a linux system you can run five computers on one box (you run linux and on that box you run a virtualization software like Vmware or Citrix or Virtual Box). Each of those computers can be different. One could be Ubuntu (a flavor of Linux which I kind of like). Another could be Windows 7. Another two could be Vista. If you are prepared to stretch the law one could even be Apple OSX. In other words one box makes five computers. Or sixty-five provided the box is powerful enough. And they share the same processing power and the same RAM which means if one person is not using it another person can.

But also Linux boxes can be used a different way. You can run ten boxes linked together and pretend they are one computer - and from the perspective of the user they are one computer. Indeed you can run a million computers together that way and they will behave like a single integrated supercomputer. We have an example - its the Googleplex in which it looks like you are sending your request to a single super-computer but the whole thing is run on desktop computers racked in huge storage barns. The beauty of the linked computers is massive redundancy. If 1000 computers went out simultaneously in the Googleplex you would not even notice. The other computers - Borg like - will just take up the slack.

Now you can do this in combo - you could run 70 Microsoft machines on two linux boxes each box being redundant. That is pretty much bombproof because linux machines barely crash and virtual machines barely crash (for reasons explained in this post).

And that is what this financial institution does. It has two largish linux boxes linked and running 70 virtualized Microsoft boxes.

And because it is a financial firm the two linux boxes are backed up second-by-second at a remote site 70km away.

The staff have their old computer sitting under their desk. They see it. They just have no idea that it is non-functional - a dumb terminal for their virtual machines with only the graphics card doing anything. (For some reason graphics cards don't yet virtualize well though that is changing...)

Given the boxes under the desk do nothing they are never going to be upgraded. The linux machines will be upgraded - but that is little more than throwing in another server blade.

Its OK for Microsoft: Microsoft is still renting 70 software licenses to run on 70 virtual machines. It is still renting office and the whole suite of other Microsoft products. But it is diabolical for Hewlett Packard who like Dell are highly dependent on corporate computing businesses for their margin.

Those businesses are stuffed. They don't exist in ten years. And the only fast-growth section is going to be server blades (see above) and those only need to run commodity linux so they will be commoditized.

Hewlett Packard is right to think there is no future in the PC business. The people who are whining at HP's actions are wrong. They should have realized there was no future in that business and sold HP stock. I never owned it (as I noted once in this post**).

Andreessen in on the board of Hewlett Packard. He thinks software will eat the world but in this case it is his company that is being eaten. He desperately wants to salvage it but salvaging it is expensive if you start from the platter Mr Hurd left him.

Anyway if pure hardware businesses are stuffed (and I think they are) then what happens to hardware-software integrated devices? If they can be replaced by software only they are stuffed (example Cisco). But it is not always that clear. Andreessen's article gives the example of military drones - pilot-less planes which can kill. The pilot is being not replaced by software but turned into a jockey with a joystick who may go to war someone in the Continental USA - killing people with drones before going back to be with his wife and kids. But the plane still exists and they still needs guns. It is a software-hardware device and it is not getting eaten. It may be getting better but the bullets are real bullets and they can't be virtualized. So they look safe enough from being eaten by software and it is the fact that the drones are lethal which makes them safe.

The point here is that software does not eat the world - it changes the world but the drone business doesn't get eaten in the way guitar tuners or Cisco routers get eaten.

So lets change the title of Andreessen's piece: software eats a good part of the world but supplements other parts - and as an investor in existing technology you need to know whether you are eaten or not.

I think I have a ready answer for this: every time you look at a piece of kit (a hardware device) you have to ask yourself whether the output of your hardware device is information or the manipulation of information or whether it is something else.

If the output of your hardware is information or the manipulation of information then you are going to get eaten. If the output is something else then you are not.

So lets do the division.

Guitar tuner: information. What is the pitch of the guitar? Doomed.

Alarm clock: What is the time? Do I need to wake up? All information. Doomed.

Military drone: Output is violent death. That is not information, it is a brutal physical reality and hence it is not eaten by software.

Cisco router: manipulation of information. Doomed.

Other items in Andreessen's article

Telephone companies. That is pure information manipulation. Ultimately doomed except for linking all this together. Certainly the old analog phone network is problematic.

Walmart distribution systems. Well the output is shopping for physical goods. Not doomed. Information is just an input.

Oil and gas exploration where computers drive drills etc: not doomed - the output is oil and gas.

Additional ones:

Libraries. Doomed.

Traffic lights. Doomed - but that will take time to get the controlled cars up.

You can go on.

For thought.






John

*We are - despite appearances on this blog - primarily long investors and we spend a lot of time thinking about our longs.

**Disclosure: I once shorted HP but made no money. The analysis was right. Our timing and execution left a lot to be desired.

The SEC should do a secondary suspension of Longtop Financial Technology

Puda Coal is a Chinese company already suspended by the Nasdaq and then relegated to the Pink Sheets. Whilst on the Pink Sheets unsupportable and probably false promotion material was circulated about them.

The SEC suspended them again - a secondary suspension from the gray market.

The SEC should do the same with Longtop Financial Technology. The company can't make payroll. The independent directors have all resigned as has the auditor. There are no independent directors and no auditor.

The company raised well over $100 million in cash. The auditor could not find it.

But we still have Weizhou Lian (the man who probably knows the location of that $100 million) running around telling everyone he has something to sell.

This is - at best - a breach of regulations on fair disclosure.

The information is either true - in which case it should be made public more generally - or it is false. In both cases the stock should be suspended - preferably before trading on Monday morning.




John

PS. I meant it when I said I no longer have a financial interest in this.

I have no position left in the stock. I write this as a public service.

Trina Solar and the meaning of "strategic partnership"

A couple of days ago Trina Solar announced a "strategic partnership" with the well managed Australian company Origin Energy*.

I think Origin is one of the best managed companies in that space in the world (but it is not exactly cheap...)

Origin also has an Investor Relations department that in my (fairly considerable) experience is dead straight: they tell the unvarnished truth and I like them.

Anyway here is the announcement:

CHANGZHOU, China Aug. 17, 2011 /PRNewswire-Asia-FirstCall/ -- Trina Solar Limited (TSL) ("Trina Solar" or the "Company"), a leading integrated manufacturer of solar photovoltaic (PV) products from ingots to modules, today announced that through its subsidiary, Trina Solar Australia Pty Ltd, it has signed a strategic partnership with Origin Energy Australia ("Origin"), the leading Australian integrated energy company. 

Under the terms of the agreement, Trina Solar is expected to supply Origin with approximately 22 MW of PV modules over the next twelve months starting from the third quarter of 2011.   

"We are delighted to initiate our relationship with Origin, Australia's leading energy retailer and the country's largest green energy retailer with significant investments in renewable energy technologies," said John Susa, Trina Solar's Country Manager of Australia and New Zealand. "We are confident that this long-term partnership with Origin will bolster our ability to expand and strengthen our market position in the residential segment." 

"Origin, which is one of Australia's leading solar retailers, has closely reviewed the capability and quality of a number of solar module suppliers in recent months in order to offer its customers quality solar solutions. The high efficiency, scale and long term strategic positioning of Trina Solar has impressed us and we look forward to a long term relationship," said Mr. Dominic Drenen, Origin's Solar and Home Products Retail Executive.

The first thing that jumped out at me was the small-scale of this relationship. 22 MWs of panels over twelve months is about 1 percent of Trina's output - and they are selling that to a major distributor in a country where the sun almost always shines, solar subsidies are still common (albeit reduced) and where the only local factory has just closed.

In other words it is not much of a "strategic partnership".

But just to make sure I wrote to the Investor Relations department of Origin.

Hi John,

Origin has not put out a release as it is simply part of our ongoing supply arrangements and not material in its own right.  Trina is one of a number of suppliers we use.

Cheers,
Angus


Angus Guthrie
Group Manager, Investor Relations

Now we can rephrase the Trina Solar press release more accurately: Trina has sold some solar panels to Australia's Origin Energy. No details as to price were announced. Trina is one of a number of suppliers used by Origin Energy. This deal is material to neither party.

Chinese Companies and Strategic Partnerships

The phrase "strategic partnership" has come to mean customer - often small customer. But it is genuinely confusing as it sounds important.

When I was looking at Longtop Financial Technology (a company now trading on the pink sheets in pennies) I came to the conclusion that much of the company did not exist. Certainly there were large claimed businesses that just could not be found.

But we were confused by Microsoft (on their website no less) listing Longtop as a "partner". You can still find the listing here. This relationship was of course hyped by Longtop but when I checked "partner" just meant "customer".

And I write this only as a warning: be very wary of any Chinese company claiming to have a "strategic partnership" when what they mean is small commercial sale. It is yet another sign of an over-promotional management team.




John

*In my past career I purchased (for clients) almost 5 percent of Origin Energy at under $2 a share. It was cheap. I sold it for just under $4 a share and thought I was clever. The stock price is now $13.72. That was not my best decision and I had full understanding of how good this company was.

Radio National program on failures of the audit profession

Radio National is the earnest national public radio service in Australia. They recently ran an hour program on the audit profession and this blog and my views particularly on China got a mention.

You can find the program here - and it is worthwhile listening.

Of course if you think that an hour talking about accountants is marginally less interesting that listening to paint dry then you might skip it along with the rest of this post.

Whatever you can find a reasonable review at Francine McKenna's excellent if polemical blog.

I appear after Carson Block just after 41 minutes... I am there as much as anything because I offer a (limited) defense of the auditors which Francine McKenna (who was given the last word) dismisses pretty harshly. The limits were edited out of the show in part because I named certain institutions as corrupt and Radio National did not want to be sued and in part because it was useful for the show to have someone (anyone) on it willing to defend what looks to be indefensible.

Here without the editing is my limited defense of auditors.

Audit is a process which is designed to catch fraud. Any defense that says that audit is not meant to catch fraud is of course nonsense.

The process involves among other things checking that a sample of transactions actually occurred and are accounted for correctly. So you look at individual transactions flowing in-and-out of the bank and check them against bank statements. You check the end balance of key items (receivables, inventory and especially cash).

If an audit is done properly (and that usually involves a competent partner and a group of competent juniors following boring rule-driven process) it will detect almost all major frauds. Competent audit may miss small scale embezzlement but will not miss wholesale fictional accounts.

The bulk of major frauds and almost all of the Chinese frauds would have been detected by proper process.

However now and again you find a really sophisticated fraud which fools that process. In China the main way of fooling the process is to have the bank in the the fraud and have the bank provide accounting statements that match the fraudulent transactions. Of course the criminal firm will also need the customers to be in on the fraud (and say to verify receivables in random checks).

In other words a fraud that fools the process needs to embed itself into a well-controlled fraudulent network. It needs to be elaborate and well constructed.

In that case - and only in that case - I am willing to offer the auditor a free pass. The auditor might have done their job properly and not detected fraud.

I think there are two Chinese frauds that meet this test. The first is China Media Express where I think it entirely possible that Deloitte did their job properly and did not detect the fraud. The other is Longtop (which was also audited by Deloitte). In both cases the customers (advertising agencies and banks) were in on the scam and the banks provided false statements to verify cash flows and cash balances.

And that is the extent of my limited defense of the auditors.




John

Google and proprietary formats

Day-to-day I use an Ubuntu computer. The problem with open source is that every now and again you need a particular codec (say a way of encoding sound) or some other proprietary piece of software protected by a patent and you just can't do something. 

This is a big problem for people distributing open source systems such as Android. Access to patents is the stated reason why Google has just purchased Motorola mobility.

But hey - as an Ubuntu user - I am getting used to this.

You see I have (now infrequent) trouble with proprietary formats with conference calls and investor presentations. Microsoft is the worst offender (but I can't expect them to change). BofA can be surprisingly hard. But most things for most companies are fairly easily accessible.

Then I tried to listen to the Google conference call for the Motorola mobility purchase using an Ubuntu 11.04 and Google Chrome. 

I can do it but I am amused by their choice of format:

Yes you do see this right. I get the choice of Flash or Windows.

Both proprietary.

John

Four days and roughly flat

After four strange days this week we have made some mistakes (including owning a lot of Bank of America), cleverly covered some shorts and bought some longs on Monday (but limited our trading because of blind fear), made a few more mistakes, had some luck (I mean really good luck) and guess what: we are flat.

We are down low single digit for the month - but hey - I reckon there are a lot of people who would swap.

We were never "threatened" meaning there was little risk of what Warren Buffett refers to as "permanent loss of capital" but it was an altogether uncomfortable experience.

But I am going to tell you something that most hedge fund managers won't admit: the difference between flat and minus seven percent on a week like this one is three parts luck and one part good judgement.

The really big negative surprise

The really big negative surprise from the week was the extent to which we were vulnerable to general hedge-fund liquidation. Our shorts - usually higher beta than our longs - and a source of most of our profits this year - just did not give us the hedge. Hedge funds were liquidating - either to meet margin calls or just to meet large anticipated redemptions. They were selling the most liquid stocks and some were buying back higher short interest stocks (hence those being conspicuously stronger than market).

In most markets our portfolio has relatively low beta (I think about 20-30 percent) meaning if the market drops 50bps we expect to be down 10-15bps and anything different is "alpha". This week we tracked market to a much larger extent and on one day (Wednesday) we actually fell slightly more than market in the first hour. I was genuinely and unpleasantly surprised.

The luck

We had two bits of luck during the week. Firstly our put-option position on Trina Solar paid off big time even though the thesis was mostly wrong.  We covered half the position. That was worth a few points. We have a changed thesis which we are very comfortable with and have positioned differently for the changed thesis.

Second we have a large (and sometimes discussed) position in News Corp. Good results sent the A class share up 18 percent. The results were driven (yet again) by Cable TV Stations. Fox News may or may not be fair and balanced. Whatever: it is extraordinarily profitable.

Logitech: another thing we got wrong

I pretty-publicly lost a fair bit of money on Bank of America. It was for a while our biggest position and whilst we trimmed some we have not played it well.

But I am just as upset about Logitech. I wrote up our basis for shorting it. I got roundly panned as an idiot on Business Insider (see the comments). I sat with the position for a while - but the results were slightly different to how I anticipated and I thought I might be wrong and covered.

Here is the six month stock chart:

Being wrong and losing money: that is part of the game.

Being right and not making money: that is really annoying.

Now I can't even remember the reason why we covered Logitech. It seems so silly... and maybe had I had the flu that day or a sprained ankle we would not have covered it. And the fund would be slightly more profitable.

Lessons

Day to day there is a lot of luck in this game. But if you avoid being stupid (owning Logitech for instance as it zoomed towards obsolescence) you will do OK over the long run.

Just make sure that there is nothing in the short run that can truly hurt you.

And be smart - really smart - but try not to be "too smart". That is a hard judgement.

Remember the difference between flat and minus seven this week is dumb luck.

And this week we were lucky and we were not too smart. Which - from the perspective of our clients - is a good thing.



J

Trina Solar: Somebody got lucky, but it was an accident

Trina Solar has become a minor obsession of mine. I was originally attracted as a fraud short as per many China names and there are plenty of red-flags. However it is not a typical Bronte China short because unlike other names we have been short (China Media Express or China Agritech) there is no doubt about the existence of Trina. Indeed a fair bit of the Trina story can be easily proved to be real.

Trina is a large company making many solar panels and many journalists have visited it.

This is not obviously Longtop Financial or China Media Express and suspected fraud is not a good basis for shorting the stock.

Still suspected fraud was our original basis for shorting the stock. We purchased a large number of put options (a scatter of maturities and strike prices) almost all of which were out of the money and all of which were effectively levered bets on bankruptcy.

We have made money on our position (fortuitously) but I am reminded of the Bob Dylan line that “somebody got lucky, but it was an accident”. Luck in this case is a market collapse and fairly good timing in our transactions. Having done considerable work I am not sure that our original thesis was correct.

We have covered much of the put position and may cover a little more but it is our intention to let most of the remaining position ride to maturity. We covered some puts by going long the stock. Put-call parity tells you this is economically similar to selling the put and buying calls – so the way to think of our current position is a “butterfly”. We make a lot of money if the stock collapses to low single digits. We make some money if the stock price moves up sharply and we lose money if the stock price goes sideways. What we have is a bearish weighted “butterfly” stock payoff schedule. Largely we are playing with "house money" now too as we have cashed quite a lot of our original position at a profit.

This (unfortunately very long) blog post goes through how we thought about Trina Solar and what we learned along the way. If you are interested in why the Chinese solar stocks trade at very low PE ratios despite massive growth rates and high ROEs then this blog post will answer your question. Ultimately though it won't tell you whether to be long or short but it will explain the risks in both positions. As I said we make money in all events except a range-bound stock – and we think a sideways stock is vanishingly unlikely.

Red flag number 1: the people at Trina Solar

What originally attracted us to Trina Solar as a short were people. Top of the list was Peter Mak – the head of the Audit Committee of Trina Solar. For those with a lurid sense of voyeurism this is Peter Mak dancing on his (I think 49th) birthday.

Peter Mak was the CFO of A-Power – a company we profitably shorted on the NASDAQ. A-Power is currently suspended with director resignations and other red-flags but the management deny it was fraudulent. I do not want to add to the debate other than to point you to Eiad Asbahi's analysis of the company.

There were other lesser red-flags at Trina Solar. For instance the current CFO (Terry Wang) used to be Executive Vice President of Finance of Spreadtrum Communications. I know nothing original about Spreadtrum but it was the subject of criticism by Muddy Waters (a firm I have found reliable elsewhere). But these are weak-tells – enough to focus your attention – but not enough to make you want to short a fast-growing company with a PE ratio of three.

Peter Mak has now resigned from the board of Trina Solar. I can't find a single solid negative thing about the new head of the audit committee (Jerome Corcoran). The only (minor) problem I have with Jerome Corcoran is that he joined the board the same day as Peter Mak which suggests some commonality of origin or purpose.

There are other red-flags too. The company used to use Crocker Coulson as an IR officer. That is a light-red flag. Crocker you see is not my favorite person as he threatened to sue me over this post on Universal Travel Group. Universal Travel is now suspended from the NYSE.

Crocker Coulson, like Peter Mak, is no longer associated with Trina Solar. 

Red Flag number 2: matching capital expenditure with capacity

The next problem we had with Trina Solar was that we found it hard to reconcile their capital expenditure with their capacity. The company has been massively expanding their capacity and that would normally take some considerable investment in machines and other infrastructure. There however appears to be little correlation between capacity investment and capacity.

Trina Solar's debt covenant for their long term debt (the covenant in my last post on this company) has the following statement about a 500 megawatt project. The project is for 500MV of ingot making and module making capacity.

Project” refers to Changzhou Trina Solar Energy Co., Ltd.’s Solar PV Industry Vertical Integration Product Project with an Annual Capacity of 500MV. The Project is located in Changzhou City of Jiangsu Province, to the south of Xinghan Electronic, the north of Nenjiang Road, the east of Chuangxin Road and the west of Keji Road, and to the north and west of Xinxi Road, the south of Nenjiang Road and the east of Keji Road. The content of the Project construction is to build a production base of solar PV industry vertical integration products with an annual capacity of 500MW and reach a production capacity of 250MW for each of mono-silicon rod and multicrystalline ingot, and 500MW for each of wafers, solar cells and PV modules. A “PV Integration Engineering Technology Research Center of Jiangsu Province” backed up by Changzhou Trina Solar Energy Co., Ltd. is to be established. The project has planned to use land in about 596 mus for construction, build an area of 329,983 square meters of new buildings and purchase production equipment (1,064 items per set). The total investment of the Project is USD597,900,000, of which the fixed assets investment is USD392,940,000, and the working capital is USD204,960,000. The sources of the Project funds are as follows: USD200,000,000 as capital funds of the Project; USD93,690,000 raised by the enterprise; USD304,210,000 coming from the syndicated loan.

Fixed asset investment for 500MW is given in the contract as USD393 million – or 78c per watt. This is somewhat more than my third party checks give me (55 cents per watt but if anyone can give me real data not sourced from a Chinese company I would be thrilled.)

Whatever – it is very difficult to reconcile this cost-of-capacity versus the company accounts.

Trina Solar had expenditure on fixed assets (gross of depreciation) of 165, 136 and 144 million for the three years ended 2008, 2009 and 2010 respectively. End of year capacity in ingots and modules was as follows:

Year	Ingots (MW capacity)	Modules (MW capacity)
2007	150	150
2008	350	350
2009	500	600
2010	750	1200

In 2008 spending 165 million purchased 200MW of capacity in both ingots and modules. That is a little more than the Changzhou project. In 2009 spending 136 purchased 250 of capacity. In 2010 spending 144 purchased 250 of capacity in ingots and 600 in capacity for modules. Quarterly capital equipment numbers are also hard to reconcile with capacity expansions quarterly.

But and this is an important but – it is entirely possible (in fact probable) that they made old capacity more efficient (by better processes etc) and so capacity spend and capacity should not entirely match. The quarterly numbers are so wonky when I started looking though I never seemed to get comfortable.

When I tried starting to get numbers to match (eg starting property plant and equipment plus new investment less depreciation less sales of PP&E etc) I could not quite get them to balance and I could not work out what I was missing. The differences were minor – and I understand that accounts never quite balance (you are always missing something) but the more I looked the more discomfort I got.

Red flag number 3: Capital Management

The biggest red-flag I had with Trina Solar was their strange capital management. Trina had – at year end – over 750 million dollars in cash – cold hard unencumbered cash on the balance sheet and was still very actively rolling money in the short term market. The cash yielded maybe 50bps. The money borrowed was maybe 8-10 times as expensive. In 2009 for instance the company borrowed $537 million in the short term market, repaid almost precisely the same amount and had substantial cash balances the whole time.

I have no problem with a company taking long term debt whilst sitting cash on the balance sheet. Indeed it makes sense to borrow money when you don't need it because one day you may need or want it and it is not available. Any company that did that sailed through the financial crisis relatively unscathed.

But taking short term debt – lots of it and actively rolling it – whilst you have lots of cash sitting on the balance sheet is unusual. Moreover the company went to market several times raising equity or convertibles – and the cumulative raising just sat in cash. Raising money when you don't need it to sit in cash always raises my eyebrow. The most telling point about Longtop's accounts was that they went to market to raise cash when they did not appear to need it. (In that case the cash raised has somehow disappeared – or at least the auditor could not find it.)

Problems with the red-flags

The problem with my red flags is that when I tested things there was always an alternative explanation. My last blog post on Trina Solar went through a covenant which I thought Trina had broken in their long term debt covenants. Indeed I thought that was a “slam dunk”. The company provided me an (entirely unconvincing) explanation of why the covenant was not broken.

However my blog readers (and you are a clever lot) managed to find the original covenant and demonstrate that the covenant was indeed not broken. What appeared to me to be a "red flag" was just a bit of the financial accounts where meaning had been lost in the translation between the covenant and the annual filing. Meaning may have literally been lost in translation from Chinese to English.

Other red-flags were like this too. Look hard enough and the red-flag disappeared.

Taking Trina's accounts seriously

My first view of Trina was to take the accounts not-so-seriously. After all some things did not reconcile and I distrusted the head of the audit committee. But as I checked things out my red-flags disappeared. Which left me pondering in an altogether different fashion. What happens if I take Trina's accounts entirely seriously – what does this say about the business?

The first thing to take seriously is the three quarters of a billion dollars in cash on the balance sheet – cash that is there despite rolling considerable short term debt. That demands an explanation.

Normal financial management would have you take the cash and pay off the short term debt. There are two circumstances where it makes sense to have cash and short term debt. The first is that the debt is in different structures to the cash. (For instance an insurance company might have debt in the parent company balance sheet and cash in the insurance subsidiary.) The second circumstance is that you might have a massive and short term need for the cash which you may or may not be disclosing.

So I went looking at the disclosure for short term debt to see if it was in different structures. Here is the key text.

Short-term borrowings

The Company's short-term bank borrowings consisted of the following

Simplified table – 2010 only	$million
Short-term borrowings guaranteed by Trina	15
Short term borrowings secured by machinery of Changzhou Energy Trina Solar Energy Corp	77
Unsecured short-term borrowings	24
Total	117

The average interest rate on short term borrowings was 7.11%, 5.14% and 4.04% per annum for the years ended December 31, 2008, 2009 and 2010, respectively. The funds borrowed under the above short-term arrangements have no covenants or restrictions, and are repayable within one year.

These numbers do not quite match the balance sheet as there is another $42 million of "current portion of long-term bank borrowings". So short term borrowings in the balance sheet total $159 million.

Then here is the disclosure that makes it all make sense. It is a doozy - but after I read it much of my original thesis about Trina just evaporated - and was replaced with another thesis. The new thesis is the main content of this blog post. So please double-read this disclosure:

The Company has short-term bank facilities of $483,851,907, $590,622,009 and $1,741,578,929 with various banks, of which $282,496,077, $327,899,446 and $650,880,259 had been drawn down and $201,355,830, $262,722,563 and $1,090,698,670 were available as of December 31, 2008, 2009 and 2010, respectively. Those short-term bank facilities are renewable annually by mutual agreement between the parties.

Trina Solar has a $1.741 billion (that is with a b) facility with "various banks" it is "short-term" and "renewable annually" and it has been drawn to $651 million.

That $651 million in debt does not appear anywhere on the balance sheet. But it is there and it is due-and -payable at some date in the next twelve months.

And now we have a perfectly reasonable explanation of why there is more than three quarters of a billion dollars in cash on the balance sheet and the company is repeatedly selling equity to raise cash and rolling over lots of short term debt.

They have a really big obligation which is "short-term" and relies on the banks to renew their facility. And the obligation is not on the balance sheet.

So I went looking for it. After all this obligation is sufficient to get Trina to willingly undertake contorted capital management. So I tried to find everything the company has said about "off-balance-sheet arrangements".  Fortunately (and a little unusually for this company) there is a remarkably simple disclosure as to the entirety of their off-balance sheet arrangements:

Other than our purchase obligations for raw materials and equipment, we have not entered into any financial guarantees or other commitments to guarantee the payment obligations of third parties. We have not entered into any derivative contracts that are indexed to our shares and classified as shareholders’ equity, or that are not reflected in our consolidated financial statements. Furthermore, we do not have any retained or contingent interest in assets transferred to an unconsolidated entity that serves as credit, liquidity or market risk support to such entity. We do not have any variable interest in any unconsolidated entity that provides financing, liquidity, market risk or credit support to us or that engages in leasing, hedging or research and development services with us.

So it is purchase arrangements. There are no other off-balance sheet obligations.

They even spell out what those forward purchase contracts are. Here is the key paragraph:

As of December 31, 2010, the Company had entered into certain long-term silicon procurement contracts, under which the Company agreed to purchase silicon materials in an aggregate amount of approximately $14.6 billion over the next four to seven years.

Like wow. Like hooley-dooley. This is the disclosure which makes Trina make sense.

You see in the last quarter revenue was only $550 million.  That is just over $2 billion a year. It is roughly $14 billion over seven years. The entire revenue line is committed to buying polysilicon.

This only makes sense if the company grows and grows and grows. Without massive growth this company can't meet its purchase obligations.

And we know what is backing the purchase obligations - lines of credit secured by bank loans and not appearing on balance sheet.

If this company stops producing lots and lots of panels (and buying lots of silicon) it goes bust because it is committed to buying the silicon.

People ask why the Chinese solar panel makers are massively expanding production even though there is a glut: well there is your answer.

Italy was a huge market for utility-scale solar plants. These require viable financial markets because a solar-farm is like a power station where you purchase all your fuel up front. They are capital intensive beasts. And if you haven't noticed the Italian financial markets are not working as well as say eight months ago.

And the subsidies are being reduced in markets like Australia and Germany.

But - driven by their humungous polysilicon purchase agreements the companies are driven to expand production. If they do not their polysilicon purchase agreements make them go bust.

If they can sell the panels at good margins they are going to make a fortune. The companies are growing very rapidly - and are compelled to do so by their polysilicon purchase contracts.

If they can't sell the panels then the companies will burn and the 750 million in cash sitting on the balance sheet will evaporate. It will be transferred to the polysilicon makers. The $650 million drawn bank loan - the one not on balance sheet - will suck them into financial oblivion.

And this contract is not easy to renegotiate because the polysilicon makers hold the bank lines which is the equivalent of holding the cash. If Trina tries to get out of the contract then the polysilicon maker just draws the bank line and gets paid anyway. And Trina will wind up owing the money to the bank...

Success or failure for Trina all depends on whether they can profitably sell the panels.

Massive and compelled growth at high margins will send this stock to $50 or more. Massive and compelled growth when you can't sell the panels at good margins will send the stock to zero. There is not much in-between - this company is attached to the rocket-ship and and will either explode or go into orbit. It has to double in size and double in size again. It may even have to double again after that. By the end of this year it will be over ten times as large as at the end of 2007 - and it is compelled to grow beyond that too.

This is (with apologies to the Clash) death or glory Chinese stock market style.

Now you can see why we at Bronte have rejigged our position so we make money at the tails of the distribution and lose money with a sideways stock. Both $50 and $0 are hugely profitable to us.  $15 in a year is ugly but given the power of this rocket-ship we don't think that is going to happen. Strapped onto a rocket-ship you are going to have a wild time.

So is it death or is it glory for Trina Solar?

Lets be blunt. At the moment it is looking like death. If nothing changes death comes within nine months, and probably far faster than that.

Production is going up, sales are going down and the difference is sitting in inventory. The company is selling some product but is also collecting its receivables much slower. The changes in the last balance sheet are a just startling - this is from the end of the first quarter.

Trina Solar Limited
Unaudited Consolidated Balance Sheet
(US dollars in thousands, except share and per share data)

		March 31				December 31				March 31
		2011				2010				2010
ASSETS
Current assets:
Cash and cash equivalents		$	489,820			$	752,748			$	636,080
Restricted cash			64,813				38,035				54,393
Investment in securities			426				296				723
Inventories			179,780				79,126				80,685
Project assets			42,110				34,979				7,196
Accounts receivable, net			542,967				377,317				305,496
Current portion of advances to suppliers			82,370				81,230				44,393
Deferred tax assets			19,903				10,258				4,653
Prepaid expenses and other current assets, net			70,394				41,149				44,159
Total current assets			1,492,583				1,415,138				1,177,778
Advances to suppliers			94,807				93,248				96,317
Property, plant and equipment, net			663,851				571,467				504,365
Prepaid land use right			36,854				37,048				27,281
Deferred tax assets			15,405				14,667				10,430
Other noncurrent assets			515				521				1,568
TOTAL ASSETS		$	2,304,015			$	2,132,089			$	1,817,739
LIABILITIES AND SHAREHOLDERS’ EQUITY
Current liabilities:
Short-term borrowings, including current portion of bank borrowings		$	153,286			$	158,652			$	221,907
Accounts payable			253,223				188,000				162,588
Amount due from related party			—				669				—
Convertible note			137,065				136,263				—
Income tax payable			46,656				34,157				12,115
Accrued expenses and other current liabilities			132,487				82,329				52,227
Total current liabilities			722,717				600,070				448,837
Long-term bank borrowings			295,652				299,977				296,102
Convertible note payable			—				—				133,838
Accrued warranty costs			44,194				38,711				24,057
Accumulated Other noncurrent liabilities			18,454				19,684				16,074
Total liabilities			1,081,017				958,442				918,908
Ordinary shares			40				40				39
Additional paid-in capital			644,628				642,830				636,747
Retained earnings			567,423				519,770				252,859
Other comprehensive income			10,707				11,007				9,186
Total shareholders’ equity			1,222,798				1,173,647				898,831
Non-controlling interest			200				—				—
TOTAL LIABILITIES AND SHAREHOLDERS’ EQUITY		$	2,304,015			$	2,132,089			$	1,817,739

The things to notice in this balance sheet is that inventory went from $79 million to $180 million in a single quarter. Accounts receivable went from $377 to $543 million.

They finally spent a lot on property, plant and equipment - probably over $100 million - as PP&E went from $571 to $664 million.

All of this is cash consumptive - and the cash balance went from $753 to $490 million. That cash fall happened whilst liabilities went up. They did not repay any debt. Cash burn was $263 million. At that rate of burn they die in less than seven months from the end of the March quarter - that is sometime about October. They might delay death by collecting the above-mentioned receivables - but that is only a short-stay of execution.

Revenue went down even though production went up (pricing was terrible). If they actually tried to sell the product to people who paid them (rather than sit in warehouses as inventory or sell it to people and count it as "receiveables") then you could just imagine how ugly the pricing would have got.

The company does not publish a quarterly cash flow statement but there was this amazing exchange in the last conference call...

Gordon Johnson - Axiom Capital Management: Just a couple of housekeeping questions, what was the operating cash flow in the quarter?
Terry Wang - CFO: It's negative for more than $100 million.
Gordon Johnson - Axiom Capital Management: Negative $100 million?
Terry Wang - CFO: It's more than $100 million, $120 million around roughly.

I figure $120 million negative is an underestimate - but without a cash flow statement that is a guess.

Bluntly though it looks like death because you can't run a business with that much negative cash flow.

This company has "profits" but negative cash flow because (in accordance with accounting standards) it counts increased receivables and increased inventory as part of its profits.

And the negative cash flow looks like it going to get worse. Much worse. You see the company pre-announced the second quarter earnings. The critical phrase is this.

While shipment volumes in the second quarter were our highest ever, sales were adversely impacted by extended slower demand and high industry inventory due in part to recently issued regulatory revisions and reduction in solar subsidies in Italy,” said Mr. Jifan Gao, Chairman and CEO of Trina Solar. “We expect a significant improvement in production costs and an increase in shipment volumes in the third quarter."

Shipments are up (320MW to 395MW in consecutive quarters) but sales are down. Guess where inventory is going? Presumably some warehouse somewhere.

Pricing has deteriorated further too - so to be blunt, cash flow has got to be getting worse. There is a possible offset if they collect the increased receivable balance described above.

Moreover they expect a significant increase in shipment volumes in the third quarter. Of course shipment volumes are going up - they have to driven by the massive polysilicon purchase obligations.

But they better hope they can sell those shipments or they are dead very rapidly. You can't bleed a couple of hundred million a quarter for very long. Not when you have to roll all that bank debt and you are obligated to issue bank guarantees over all those purchases.

And given the big markets for this are in Southern Europe (where solar-farms are suddenly hard to finance) the chance of selling all that product is reduced.

So I think the outcome is death and our position in the stock is sharply weighted towards death.

The glory-case for Trina solar

Fast growth into declining sales does not necessarily mean death if sales miraculously turn around. A large Chinese solar subsidy might do that but the recently announced solar subsidy is set at a low level (feed in tariffs are under a third of the European equivalents). The really rich markets (Southern Europe) don't look like they are coming back soon.

The second - and more important way that they can sell the massively increased production is if they cut prices far enough that the relatively expensive capital markets of Europe can finance the projects. They will do that if they get (even-more) super-efficient and if the polysilicon costs drop far enough.

They have been very efficient in the past - processing costs have dropped but nothing like as fast prices for panels.

But the real cost drops have been driven by polysilicon which peaked at over $400 per kilogram and is down by at least 80 percent.

The main polysilicon contract is attached to the 2009 annual report. This contract specifies the way in which polysilicon prices in the contract are to drop if the spot price for polysilicon drops.

In other words they are obligated to buy lots and lots of polysilicon - they are just not obligated to buy it at the current price.

The contractual terms of their lending agreements require that they do actually process the polysilicon.

In other words if they can get their processing costs low enough and they can grow the market enough, they might be able to sell all of their obligated polysilicon by growing the market.

If this happens the company - now on a PE of 3 - will (at least) quadruple in size and get repriced as a growth stock. $50 in that case is conservative. Glory for this company is a long way up from here. A very long way up from here.

But they better start getting the costs down and selling the product really rapidly (like right now) because the current large and increasing negative cash flow will leave this company as just more road kill in the US listed Chinese space.

For thought and conversation.





John

Paul Krugman vs Mr Bean

Welcome to the newer readers from Paul Krugman's blog. It is the third time (I think) he has linked me and as per usual he brings a lot of traffic.

My post on margin calls is my second most visited post in the past year.

The most visited post however is Mr Bean declaring the European Debt Crisis over.

Krugman is super-smart, has a platform at the New York Times and at Princeton and has a Nobel prize.

But Rowan Atkinson pulls a bigger crowd and doesn't have to say anything.


John

Who has got the margin call?

Bronte Capital tends - as this blog has said many times - to be long large cap and relatively liquid companies at 10-16 PE ratios.

And we are short the wildest range of frauds, fads and failures you can imagine.

Our shorts are small and relatively illiquid. They are also higher beta - so we tend to be 2-2.5 times as long as we are short and that is only mildly market correlated.

The strategy has been wildly profitable for the last six months and is not bad most the time but we do not like days when the market really pukes.

When the market really pukes it is the biggest, most liquid names that get sold. Why? Because you can sell them.

And they go down hardest.  Which is problematic if you are long that which goes down hardest.

So today is not our strongest day - and whilst we are down far-less than market - I can't say I am enjoying it.

But I am fascinated.

You see the desperate selling of the biggest liquid names is a sign of margin calls.

The market is not puking. Some prime broker is puking the stocks held by one or more very large hedge funds.

So lets play the game: guess who got the margin call!

Guesses by email or in the comments...



John