The consensus reaction to my last post was that Longtop does not actually provide a cloud based storage service at the petabyte level despite the plain reading of the press release. Readers have argued that what Longtop is saying is that it helps customers (who are largely financial institutions) set up this storage on their own premises using cloud technology. Longtop is thus a system integrator for the bank and probably installs third party equipment for the bank.
Moreover the consensus was that no financial institution customer has a petabyte of data they need full access to. One reader cited this fact: that in 2008 Yahoo claimed the world's largest and most active database at 2 petabyte. This article aslo suggests that the IRS data-mining database is a svelte 0.15 petabyte.
A petabyte of data is possible with extensive video and photo storage (people cited cameras) but whilst banks have some security cameras (eg on ATMs) they generally do not require this data to be stored huge lengths of time or be super-accessible.
Obviously the really big databases (eg Facebook with all those photos) are substantially larger than any needed by a bank or the IRS – but the “petabyte” claim was almost certainly marketing puff.
Marketing puff is not that uncommon in tech-stock land. The geeks even have a word for it: vaporware. Typically however vaporware refers to software that is announced without a release date: in this case it was a vapor-launch (the announcement I quoted was an “official launch”).
But puffery is puffery and does not challenge the fundamental veracity of the accounts. And indeed I have not challenged them (unlike Citron). Instead I just look at wonder at how a business can be that capital efficient.
So how exactly does Longtop use its capital?
How Longtop uses its capital is an important question because Longtop raised 127 million in late 2009 and – according to their accounts – they had no use for the cash. Maybe they planned a big acquisition but they never did one large enough to make a serious dent in their cash hoard.
Longtop currently carries – relative to its expense base which is the right way to measure it – six times as much cash as Microsoft. I went through the numbers in my first post. Longtop is a cash generation machine.
But not only is Longtop a cash generation machine it does not seem to need capital to grow. Its incremental capital efficiency is breathtaking and becoming increasingly so.
Longtop incremental capital efficiency
In the 20F (annual filing) covering March 2009 to March 2010 the company grew nicely. Revenue rose from 106.3 million to 169.1 million – a China-like 59 percent per annum. There had been an acquisition in the year ended March 2009 so the organic growth rate was somewhat smaller but still very large.
Here is the the fixed asset summary from the 20F for the year ended March 2010.
| Fixed assets, net (numbers in USD thousands) |
| March 31 | ||||||||
| 2009 | 2010 | |||||||
Equipment and fixtures | $ | 9,654 | $ | 13,022 | ||||
Leasehold improvements | 2,040 | 1,718 | ||||||
Buildings and renovations | 222 | 19,687 | ||||||
Motor vehicles | 1,253 | 1,355 | ||||||
| 13,169 | 35,782 | |||||||
Accumulated depreciation | (7,316 | ) | (9,254 | ) | ||||
Impairment | (185 | ) | (185 | ) | ||||
| 5,668 | 26,343 | |||||||
Construction in progress | 9,190 | — | ||||||
Fixed assets, net | $ | 14,858 | $ | 26,343 | ||||
| Equipment held under capital leases had a net book value of $1,292 and $732 at March 31, 2009 and 2010, respectively. |
| Construction in progress at March 31, 2009 consisted of an office building and renovations which were under construction and not ready for use. |
| Total depreciation expense recognized in the years ended March 31, 2008, 2009 and 2010 was $1,780, $2,808 and $3,193, respectively. |
Note that equipment and fixtures rose from $9.65 million to $13.02 million - an increase of 3.37 million. In that time the staff numbers went from 2602 to 4258 - an increase of 1656 employees (63 percent growth).
I want to observe something: the equipment and fixtures - before depreciation - rose by only two thousand dollars per employee.
Lets spell this out: this is a world beating software development firm with world-class economics and enormously fat margins. By its own admission it is critically dependent on the research and development done by its staff. And the incremental capital spend per new staff member would buy good desktop computer and a cheap desk and chair. Given things like power protection, backup servers etc are included in this additional fittings and equipment ($2000 per incremental employee) we can safely conclude that the new employees are treated skint. Very skint.
Whatever, there are no in-house restaurants, basketball courts, table-tennis tables and other splurges on new staff. This is not Silicon Valley. There are probably not even incremental sophisticated computers for them to test their programs on. (And they are writing software for complex environments and things need to be tested...)
I always thought it was hyperbole when Warren Buffett praised Jack Ringwalt for being late to a meeting because he was driving around trying to find a parking meter with unexpired time. But these guys make Ringwalt look like a spendthrift drunk.
This reluctance to spend is - well - amazing. Especially as there is no shortage of cash.
The tight rein on capital expenditure is even more amazing this year
From the most recent 6K revenue is up more than 40 percent since last year. Indeed the revenue rise for the first nine months is about the same as the revenue rise for the whole previous year. They have not told us how many more staff they have employed but unless the staff have become massively more productive they have probably added a further 1600 or more staff. (That rise would also be consistent with their rising cost base.)
But the total fixed assets (including vehicles, buildings, leasehold improvements etc) have risen from 26.3 million dollars to 27.9 million dollars. If we guess 1600 staff as above (and that is just a guess) then the rise is only a thousand dollars per staff member. Whatever - we are now talking cheap computers and no vehicles, fittings, fixtures or anything else much.
Even if there were no new staff it would be pretty amazing to grow fixed assets by only 1.6 million dollars whilst growing fat-margin revenue by about 70 million per annum. I have never known any business to have that sort of capital efficiency.
What is going on with capital management at Longtop?
The capital management here is just strange. The company raises money when it doesn't need it. It sits on cash equivalent to six years of operating expenses (which is about 6 times more than Microsoft). It has a world-beating software business which they are justifiably growing as fast as they possibly can. They state repeatedly that they are dependent on the skills and product development of their staff.
And despite this they don't seem to spend anything on fixed assets to make the staff more efficient or more happy.
I guess Chinese workers just put up with it. Chinese software workers make do working on outdated equipment. And as for the basketball courts and in-house restaurants of Mountain View California. Well - you ain't seeing them here.
I don't get this. Obviously I don't understand China. A riddle wrapped in a mystery inside an enigma you might say - but that was said about a previous superpower.
I just report it. Can't say I understand it.
John
24 comments:
John,
I can't say I see into SW industry that much anymore, but I know that a bunch of angel/PE investor friend of mine still love it for exactly the thing you talk about - capital efficiency. You can run multiM SW company with zilch fixed assets - you can even lease the computers you develop on if you're really into it (which can even be a better deal, due to the extremely fast obsoleteness).
You generally need quite a bit of cash outlays (most of it spent on salaries), but once you have an installed client base that is regularly upgrading you can even end up with nice positive free cash.
I've seen a very successfull business that was effectively hiring grads to do the coding, paying them peanuts for top quality work - because then they could put work experience there on their CV and after 2-3 years of "apprenticeship" get 3-5x their original salary within week of leaving.
Anon 12:30
OK, your explanation sounds great, but then why the rush to raise all that capital?
'I've seen a very successfull business that was effectively hiring grads to do the coding, paying them peanuts for top quality work - because then they could put work experience there on their CV and after 2-3 years of "apprenticeship" get 3-5x their original salary within week of leaving.'
That happens in SV because you can leave FB/twit/goog/Gpon and get that multiplier after 3-4 years *if* you are good at your job.
There's no similar model in China.
I guess Citron's explanation is they are hiding all of their costs off balance sheet, which would make sense relative to your criticism. But one would think that would show up on the cash flow statement, which I am not seeing when I look at their latest release at least. I guess Morgan Stanley's contention is that their margins are inflated because they pay their employees more in stock. The disclosure below may add some credence to that?
"If we had included share-based compensation expenses in our Non-GAAP Adjusted Net Income for the nine months ended December 31,2010, Adjusted Net Income would have been US$94.2 million lower or Adjusted Net Loss of US$15.0 million for the nine months ended December 31, 2010, and our Adjusted Net Income margin would have been negative."
Not that this is a positive, just trying to understand it same as you present.
While looking at this company I am reminded of that synical principal:
"If you are going to tell a lie, make sure that it is a big one!"
John
Peta-byte systems dont have to exist entirely within the Database, cheque image archives for instance use very large storage capicity which have comparatively small databases that contain the indexing . First Data Corp for instance have 30 petabytes of cheque records in their archive...the really large systems are the big image archives that contain images of business records (contracts, cheques, transactions) so the reference maybe to either building a large archive system (all the banks have them) or possibly a Teradata installation System 12 which have a capacity of 50 P compressed.
Having a look a their site though does raise eyebrows, the dont list any customers by name? it not unusual to have say 50% of your customer wins by reference "Longtop wins contract with a big 4 bank in China" its very unusual though to never (well not in the last few years) announce a named customer.
They are also not in the
On capital efficiency, if you compare asset turnover of US and Chinese peers, they are in line at about 6.0 - 7.0x. Check ADVS US (look all the way back to 1995), VIT US and 268 HK which are all comparable in some ways, and you will see that all of them have generated proportional revenues from their asset base.
Longtop's profitability is a different issue which you haven't raised but is justifiable given their current actions, but will have to come down over time.
On your point on capital efficiency, if you compare asset turnover of peers in the US and China they are comparable at about 6 - 7x. Advent, VanceInfo and Kingdee all generate proportional revenues for their fixed asset base, and so I don't think Longtop's fixed asset base is out of whack.
Do you understand that you keep citing sizes of SQL databases while the fluff piece explicitly said this is not what they are using? ("As a result, it can overcome the inherent deficiencies of traditional Relational Database Management System products...") You are comparing apples and porridge.
Obviously the really big databases (eg Facebook with all those photos)
Facebook doesn't really have that many more photos than Mastercard has transactions, does it? The photographs themselves take up a lot of memory, but that shouldn't really be counted as part of the size of the database - a library doesn't need to get bigger index cards if it buys bigger books.
The facebook photos have to be stored somewhere that can be recovered. The Yahoo database required roughly 1000 pcs to run.
The amount of storage you can attach to a single PC is finite.
--
No I think the Facebook database is MUCH more complicated as well as larger.
And the question was about whether the Petabyte claim was absurd. I think it is.
The second question was whether the it mattered whether the petabyte claim was absurd (or whether it was marketing puffery). I am not sure about the answer to that.
And my readers identified the second question as the key one - the one I could not answer.
For that I thank them.
John
A long comment I wrote just got FUBARed by blogger.
The summary:
The puffery doesn't matter since no one is ever going to ask them to run a PB in the cloud.
The capital lease payments are too small to account for any meaningful equipment purchases - perhaps another $1k per employee.
The stock option payments are too high nearly $200k per employee.
So the "unrelated" HR company must be providing both human and some fixed assets to allow the human to do some programming.
And the HR company is probably getting paid in stock somehow.
John,
This is pretty how most software outsourcing companies work at their growth stage.
They are not a product company who need to invest in superstar programmers. They need bodies who can do basic coding, especially given they service Financial Services clients, most of the work will be either COBOL Mianframe or AS/400. Mostly maintenance work which doesn't require complicated machines.
For the talk about cloud computing etc enterprises still run stuff inhouse and outsource support and maintenance.
And $1000 sounds about right for a new employee. That's what we used to to budget for a getting a new hire on board for outsourcing contracts. The space is usually leased and in some cases provided by the local govt.
The last two articles about LFT are just absurd on most topics. Do you have any experience on SW development? I used to like to read your analysis on your web site. But these two are really disappointing
.
He might be publishing these bits and pieces to learn from commentors who have actual relevant experience.
For me it is clear that John is stating that based on his experience the numbers don't make sense. And that experience is on financial statement analysis. As a good analyst you can draw three conclusions:
1) Numbers are wrong e.g. fraud
2) Numbers are right e.g. miracle
3) You as an analyst don't understand the underlying business, therefore your conclusions are not meaningful
You can close the last point out by speaking with people who actually understand the business and testing your hypothesis with them.
Actually, the business I wrote about wasn't in SV but Eastern Europe (and they were paying peanuts even by EE standards).
Infosys in India wasn't exactly paying peanuts, but if they continue to be India's "IT university" (Infosys trains them, competition poaches them), they as well might start doing it.
As for why they raised the capital - well, here's a scenario:
- we know they like to sit on a lot of cash (they had over 200m in account even before they raised the cash).
- they wanted to do an acquisition (in fact, they did 2010-03, and it cost them 70m). They wanted to do it with stock, but couldn't for whatever reasons - so instead they sold stock and used the cash for acquisiton.
- they over-estimated how much they would have to spend on the acquisition - yes, I know they would likely know the price before they extra capital raising, but it's hard (and expensive) to stop a running train, plus we know they like cash, so why not?
What I'd like to know is why they like to hoard this much cash over long term? Sure, having a nice cushion is good - but not having two+ years of it unless you expect to grow very very quickly.
If the management was to go and defraud it, it could have done so by now (say by doing misvalued small acquisitions). That of course doesn't mean they don't have something fancier up their sleeves, but on the evidence so far I'd say it's not obvious.
The short answer to that anon is they a haven't been audited yet. The HR company is like s siphone, it makes what should be apparent form an accounting perspective opaque. Companies who do this kind of thing have a bad track history when the covers are lifted, these kinds of structures are put in place for a reason....that is usually to stop transparency....the question is what are they hiding, the margins are impossible....not just high but completely impossible
Longtop has been gamely refuting assertions that have been put forth publicly by some short sellers and appears to have made a priority of open communication with its investors. This company's management with its openness, not evasiveness, is strikingly unlike the other Chinese companies that Bronte and others have targeted with accusations of fraud to great success, and this behavioral difference leads me to suspect that the short sellers may be misguided about this one. I don't claim to know what the truth is, but if I were a gambling man, which I am, I would be inclined to short the short sellers in this case. In short, I think it's possible that the critics are overreaching or grasping at straws.
Mr. Hempton, I've been a reader of this blog beginning a month or two after its inception and am a great fan of yours. I read everything you write and have been inspired by your investment ideas on a number of occasions. I rarely post publicly, so I thought I would take this occasion to let you know how much I enjoy your blog: long may it live!
P.S. Since Paul Krugman has referred to your blog on his, I've wondered on occasion if you and he ever connected directly, or if he's been too aloof.
There is less than 1% chance that there is not something rotten. Companies simply dont put forward these types of structures and ridiculous returns if they are real. Its like claims of unaided levitation , they can be as bold and open as they want, they can show is pictures but physics says they can float above the ground unassisted . You dont need anymore proof than the accounts, all those massive margins and the majority of transactions opaque behind a related party. They claim its not a related party despite having common signatories and a common address.
Who do you think you have to know to get contracts with the four state policy banks in China to do this type of work? Probably nothing less than Wen JiaBao, and I'm not exaggerating or using hyperbole.
This isn't to say the Company can't do the work, but so could others and for cheaper. There is simply an unidentified intangible asset here. If the Company is wise, it will use this money to build a real company, which to some extent it clearly it has. It will also take the opportunity to raise capital at a time where investors believe long-term returns are capital are higher than they probably will be.
You can look under the rocks of capex per employee all you want -- but 2-3k vs 5-6k doesn't seem like it would move the needle in terms of capital costs for their growth. Real estate office rents in China are dirt cheap and there are plenty of empty office buildings to occupy.
My opinion is that this simply a case of getting very favorable contracts.
Gotta love the most recent press release.
"We just signed a big new contract! With a major bank! No, we won't tell you who it is or any sort of expected revenue, why do you ask?"
It's press releases like these that make me more confident as a short.
I manage a company in China and have Chinese accountants with plenty of experience.
If asked, they would state very clearly that Chinese companies do not have trust worthy accounting practices, as a rule. Simple as that.
Good Call on Longtop - I'm sure you've seen the latest ADR news.
Longtop raised 127 million
Post a Comment