Inside the Kingdee bubble
18 March 2019
On the face of it, Kingdee International Software Group Co Ltd (Kingdee, 0268) might seem like a made-in-Shenzhen success story, a demonstration of what its Chairman, CEO and largest shareholder Robert Xu Shao Chun (Mr Xu) calls "the Chinese management model". Who knew that there is a better way of doing things, unique to China? The stock has more than tripled in the last 2 years.
Dig deeper, and you will find that it's a bubble stock, in a company which has relied on sector-specific tax breaks, government grants, property investment gains and questionable transactions with related parties to book any profit at all. Pour yourself a coffee and we'll explain.
At the closing price of $10.62 on Friday (15-Mar-2019), Kingdee has a market value of HK$35.08bn. The 2018 results published last week show revenues of CNY2.81bn (HK$3.20bn at the year-end rate), so it is trading on 11.0x revenues. You read that right: revenues, not earnings. We'll get to whether there is any E in the P/E later.
Note 4 of the results shows an alarming increase in the "loans to related parties", up from CNY164m to CNY722m. All of this increase happened in the second half of the year, and without any announcement of new connected transactions, if these are. The 30-Jun-2018 interim report showed a balance of CNY134m, because CNY30m had been repaid in the first half. At that point, these loans were to 2 companies controlled by Mr Xu but formerly controlled by Kingdee. The results do not include any explanation for the increase, or any discussion of the balance sheet at all.
2016 disposal: off balance-sheet, but not off-risk
To understand the origin of these loans, we need to review a transaction announced on 28-Jul-2016, in which Kingdee sold interests in 3 companies to Mr Xu:
Shenzhen Qianhai Baidi Network Co Ltd, "principally engaged in the provision of courier and logistics information services such as courier enquiry and tracking services". Kingdee sold 100% for CNY55.7m.
Shenzhen CloudHub Network Co Ltd (Cloudhub), "principally engaged in the provision of mobile Internet enterprise services through a mobile office platform". Kingdee sold 85% for CNY43.3m and retained 15%, subsequently diluted.
Shanghai Kingdee Medical Software Co., Ltd. (SKM), "principally engaged in the development and operation of medical information systems and software and medical records database, as well as the provision of mobile Internet hospital services through mobile service platforms which connect patients and hospitals." Kingdee sold 100% for CNY8.4m.
All 3 were loss-making. No figures for their respective net assets/liabilities were disclosed, but for the total price of CNY107.4m, Kingdee booked a disposal gain of CNY138.8m before tax (CNY133.6m after tax), implying that the 3 had combined negative net assets of CNY31.4m.
The stated reasons for the disposals included "to alleviate further operating margin and financial pressures on the Group". It surely did, and we'll get to that below. However, it didn't eliminate financial risk, because Kingdee generously left outstanding loans to CloudHub and SKM in the amounts of CNY61.7m and CNY105.0m respectively. No reason was given for that. The 3-year loan agreements purportedly pre-dated the sale, being dated 1-Jan-2016, which was a public holiday even in mainland China. On 14-Dec-2018, Kingdee announced that the loans would be extended for another 3 years, but with the Cloudhub loan reduced by CNY30m. Giving reasons for the extension, Kingdee described the two borrowers as "important business partners of the Group" to which it should "continue to provide financial support".
There was no discussion in the announcement of how Mr Xu would divide his time and efforts between his private and public entities. In each case in this article, under the Listing Rules thresholds which were relaxed some years ago, none of the connected transactions was large enough to require independent shareholders' approval, so it was left to the board of directors (excluding Mr Xu) to approve them on each occasion.
Going back further, we see that SKM was set up by Kingdee in 2011 to buy 75% of Guangzhou Wisetop Computer Co Ltd, which it renamed "Kingdee Medical Software Technology Co Ltd" (KMS). The price on that deal, announced on 1-Aug-2011, was CNY120m, of which about CNY46m was subject to adjustment based on 2011 and 2012 profits. Kingdee, via SKM, then injected CNY30m of cash and CNY26m of existing business into KMS to raise its stake to 81.48%. So the outstanding loan of Kingdee to SKM basically represents the high price it paid out to the unnamed vendor of KMS.
The 2011 Kingdee accounts (Note 35(a)) show that KMS made full-year turnover of CNY38.98m and net profit of CNY4.13m, so the P/E on the purchase was about 29. The 2012 accounts (Note 34) show that as a result of the (poor) performance of KMS in 2012, the contingent consideration was reduced by CNY13.01m, recognised as a 1-off gain in Kingdee's income statement. The 2013 accounts (Note 21(c)) show a further reduction of the contingent consideration by CNY23.16m, generating another gain. Despite this, in both years there was no impairment of the goodwill previously paid on the acquisition of KMS. Isn't it great how poor performance of a subsidiary can increase your profits? In the following years there never was any impairment of the goodwill on KMS, but the 2016 disposal announcement reveals that in 2014 and 2015, SKM (including KMS) made losses of CNY7.57m and CNY18.18m respectively.
Now we turn to Cloudhub. The 2016 disposal announcement stated that in 2014 and 2015, Cloudhub lost CNY24.1m and CNY81.3m respectively.
Reading the Kingdee reports for 2016 and 2017, you might forget that Kingdee had sold 85% of the Cloudhub business. The word "cloud" appears 123 and 151 times in the 2016 and 2017 annual reports. In the 2016 annual report, the Chairman writes "the business volume of CloudHub experienced robust growth." Well he should know - he owned it.
Meanwhile in an unannounced transaction on 29-Feb-2016, Kingdee bought 58% of Shanghai Guanyi Cloudcomputing Software Co Ltd (Guanyi) for CNY247.4m including CNY191.9m in cash and CNY55.4m contingent payout, although Kingdee reckoned on reclaiming CNY16.7m under a profit guarantee. The 2016 accounts detail the transaction in Note 37 but fail to disclose what the turnover or profit contribution of Guanyi was post-acquisition or what it would have been for the whole year. Such disclosure is required by IFRS3 (or its HK equivalent, HKFRS3, para B64(q)). In 2017, the related profit guarantee was cancelled, and Kingdee booked a gain of CNY25.1m, implying that the final earn-out was only CNY13.6m. Kingdee still holds CNY181.3m of goodwill for this acquisition, unimpaired.
The return of Cloudhub
On 6-Mar-2019, a week before the results announcement, Kingdee announced that it would buy back control of Cloudhub from Mr Xu, paying CNY50.5m for his 51.73% stake. The stakes of both Kingdee and Mr Xu had been diluted since the 2016 transaction, reducing Kingdee to 9.13%. The other 39.14% is said to be owned by unnamed independent third parties.
The announcement revealed that in 2017 and 2018, Cloudhub lost CNY140.6m and CNY122.5m respectively, and had negative net assets at 31-Dec-2018 of CNY326.4m. We don't know what it lost in 2016, but if we average between 2015 and 2017, then that would be a 2016 loss of CNY111m, of which (pro rata) about CNY46m was in the 5 months after the disposal. Hence we estimate Cloudhub has lost about CNY309m from its disposal in 2016 until 31-Dec-2018, plus whatever it has lost since then.
Now, we don't (yet) know which related parties increased their borrowings from Kingdee in the second half of 2018 and whether that includes Cloudhub, but whatever outstanding loans there are to Cloudhub will disappear in the Kingdee group accounts when Cloudhub becomes a subsidiary again. Magic!
While Cloudhub was controlled by Mr Xu, there were ongoing connected transactions between Kingdee and Cloudhub in the form of distribution by Kingdee of Cloudhub's products and cross-promotion of Kingdee's products. These transactions were renewed as announced on 2-Jan-2019, and it seems fairly clear that Cloudhub is and always has been an integral part of the Kingdee product offering and strategy - see for example this page on the Kingdee web site. In the 2017 annual report, the Chairman wrote: "the newly-launched CloudHub was further integrated with the Group's Cloud products, including Kingdee Cloud, Jingdou Cloud and Guanyi Cloud...".
In the period from disposal on 28-Jul-2016 to 30-Nov-2018, Cloudhub paid Kingdee fees of CNY37.3m for promoting its "PublicCloud Products", and in the other direction, Kingdee paid Cloudhub CNY61.0m as a distributor of Cloudhub's products and for promoting Kingdee products.
All of this invites the inference that Cloudhub was used as an off-balance-sheet part of the Kingdee group, helping absorb losses on an integral part of the Kingdee product offering while receiving financial support from Kingdee, and is now brought back without those losses ever hitting the Kingdee income statement; instead there will be another large goodwill item representing the difference between the purchase price and the negative net assets of Cloudhub. We hope that their software engineering is as good as their financial engineering.
Since 2008, Kingdee has had a growing property investment segment which appears to be surplus floors or buildings in its Shanghai and Shenzhen "Research Centers". We say "appears", because the last time there was even the vaguest description of these properties was in Note 9 of the 2012 annual accounts, but even then, there was no disclosure of the exact addresses as required by paragraph 23 of Appendix 16 of the Listing Rules when your investment properties exceed 5% of total assets. At 31-Dec-2018, the investment properties were valued at CNY1.82bn (HK$2.07bn), or about 40% of the net tangible assets attributable to owners of CNY4.57bn (HK$5.21bn).
In the 11 years since 2008 inclusive, Kingdee has booked gains on revaluation of its investment properties of CNY590m and net rental income before tax of CNY453m. There's nothing wrong with the accounting on that, but it does help mask the poor profitability of its software business.
We combed through the management discussion and analyses since 2008 and could not find any mention of the strategy or plan for this business, presumably because there isn't one. No discussion of tenancies, rental yields or occupancy rates, for example. The Kingdee segment analyses simply record it as "Others", even though it contributed CNY150.7m, or 36.6%, of the 2018 profit attributable to shareholders.
Corporate income tax break
Kingdee's main subsidiary, Kingdee Software (China) Co Ltd, is what China calls a "National Important Software Enterprise" (NISE), so it gets a preferential tax rate of 10%. NISE if you can get it. We can't blame Kingdee for eating a free lunch, but it does underline the central planning and sector-favouritism that goes into the PRC economy, and we've seen the same policies encroaching on Hong Kong. The question is whether sector-specific tax rates are sustainable if China decides, as part of its trade negotiations, to stop favouring specific sectors.
Even in the property investment ("Other") segment, Kingdee applies this 10% rate to its revaluation gains and net rental income, although it is difficult to see why it should be treated differently to other landlords. It probably holds these assets through its NISE-qualified subsidiaries because Kingdee staff are in other floors or buildings of the same sites.
VAT tax break
Under PRC Value-Added Tax (VAT) regulations, companies involved in developing software and integrated circuits in the PRC collect VAT at 16% (before 1-May-2018: 17%) on their sales but then get a refund which reduces their VAT (net of VAT they have paid on input costs) to 3% of their sales. This has been a huge contributor to Kingdee's profitability. In 2018, Kingdee booked a VAT refund of CNY178.5m, or 39.8% of its pre-tax profit. In the 11 years since 2008 inclusive, it has booked CNY1.60bn of VAT refunds. This again may be an unsustainable state subsidy. If the refund went away without any rate adjustment, then customers would still be paying the 16% VAT and if they are VAT-registered enterprises themselves, they would still get the input credit for it. So it is hard to see how Kingdee could raise its prices to compensate for the lost refunds.
As if the 10% income tax rate and the VAT refunds weren't enough, the state also pours government grants into software developers. Kingdee booked CNY43.3m in 2018 and CNY491m in the 11 years since 2008 inclusive. Again, we can't blame Kingdee for eating a free lunch, but that stream of Chinese take-aways is probably not something to depend on, particularly if China is trying to demonstrate to its trade partners that it doesn't micro-manage the economy or subsidise its corporates.
We've mentioned several of the one-off gains which have supplemented Kingdee's "operating profit" over the years, including the 2016 gain on disposal of Cloudhub and 2 other entities, and the gains on reducing deferred acquisition pay-outs without impairing goodwill. Here's a summary of all of them since 2008:
Kingdee 1-time gains
The core software business
OK, so despite all of that, excluding all the one-off gains, surely the market can't be wrong? Of course it can. Here are the combined core operating profits of Kingdee and, assuming they had never shifted it out-and-back, Cloudhub, starting with the reported "operating profit". As noted above, 2016 is an estimate of 5 months of Cloudhub's losses:
Kingdee core profit
So now you see that even with VAT refunds, the software business struggles to make much profit. After deducting VAT refunds, it made operating losses in 6 of the last 8 years. If you also factor Cloudhub back in, as if they had never shifted it out, then you are looking at a combined operating loss of CNY65m in the last financial year and CNY950m over 11 years. In that time, they've made more investing in property than they've lost on selling software and services.
Yes, revenue has grown, by about 38.9% between 2011 and 2018, or a compound average of 4.8% per year - but that's only a bit more than consumer price inflation, even if the growth rate has been higher in the last 2 years. Investors should not extrapolate that erratic record too far.
Funding and shareholders
Mr Xu currently holds about 24.1% of Kingdee.
Kingdee has conducted 2 substantial fund-raisings, in 2014 and 2015. In 2014, Kingdee issued a US$175m (HK$1357m) bond convertible at HK$3.90 per share, resulting in 348.075m new shares when it had fully converted on 24-Apr-2018. On 16-May-2015, Nasdaq-listed Chinese online retailer JD.com Inc subscribed for 288.526m shares at $4.60, raising HK$1327m gross. That was 10% of Kingdee at the time, although subsequent dilution would make it 8.74% unless it has bought more shares. There has been no subsequent filing. Amongst institutions, Capital Group holds 10.08%.
Kingdee has been quite popular on the Shenzhen Southbound Connect, where the depository currently holds 8.61% of the stock, down from a peak of 15.03% on 28-May-2018. The outstanding aggregate short position disclosed to the SFC was 7.77% at 8-Mar-2019, making it the 5th most shorted stock as a percentage of outstanding shares, excluding H-shares and ETFs. That also means there are 107.77% long positions.
Other corporate governance issues
Apart from the unexplained increase in "loans to related parties", the balance sheet holds some other horrors. Given that the bonds fully converted and with all that money from JD.com, Kingdee has far more capital than it needs. It hasn't invested in data centers or other capital-intensive infrastructure - is outsources that, including to Amazon's cloud. It has fixed assets (excluding the investment properties) of just CNY558m. It has net cash of CNY1511m, and it has financial assets of CNY639m which are probably all "wealth management products" based on the interim report, although we'll have to wait for the 2018 annual report to be sure. The cash flow statement shows a heavy turnover of these - they bought and sold or redeemed over CNY3.7bn of financial assets during the year.
In 2017, Kingdee also seems to have started a money-lending business, with year-end loans to third parties of CNY65m, growing to CNY108m in 2018. The 2018 cash flow statement refers to this as "the micro-credit business".
Despite being listed in HK and having no listing anywhere else, Kingdee holds all its AGMs in Shenzhen, making attendance challenging, particularly if you don't have a Mainland Travel Permit or a visa. We suspect non-employee attendance is minimal. AGM locations should not be chosen for employees who can see their bosses any day that they are available - they are an annual opportunity for outside shareholders to meet with all the directors. A review of the Corporate Governance reports shows that none of the NEDs or INEDs attended the AGM since reports of attendance began in 2013, in breach of then Corporate Governance code provision A.6.7, even though Kingdee claims to have conformed with the Code.
Despite the huge CNY2.15bn (HK$2.45bn) pile of net cash and wealth management products, Kingdee has proposed a dividend of just CNY0.01 for the 2018 year, or about CNY33m in total. How they reach that number, or why they even bother to pay such a token amount, is a mystery. There is no mention of the obese balance sheet in the Management Discussion and Analysis (MD&A) in the results announcement, nor has it featured in MD&As in previous annual reports, in breach of 32(1) and (2) of Appendix 16 of the Listing Rules. Readers would be forgiven for thinking that none of the directors had actually read the balance sheet, and also that the for-profit Stock Exchange of Hong Kong has been failing to enforce its own Listing Rules on disclosure, because they don't want to put more resources into actually reviewing the disclosures.
So what's it worth then?
Enterprise software is a competitive market, but there's clearly a real business in Kingdee - it just rarely or barely makes a core operating profit, is heavily dependent on state favours to its sector, and is run by people who appear to be comfortable with out-and-back transactions over Cloudhub, lending to businesses controlled by the Chairman, investing shareholders' money in wealth management products and starting up a microfinance business; and uncomfortable with the idea of independent shareholders talking to independent directors at AGMs and perhaps asking awkward questions like "where's the return on my equity?" It's difficult to attach much value to such a business and if this is "the Chinese management model" then we'll manage without it, thank you.
Kingdee has a net tangible asset value of HK$1.58 per share including $0.74 of net cash and investments and $0.63 of investment properties, so that provides some modest underpinning. Although Mr Xu calls himself the "controlling shareholder", his 24% stake is not enough to prevent activists pushing for change after the bubble bursts (as all bubbles do), so some of that value could be extracted, but as the stock is trading at 6.7x book value, at this point it doesn't matter.
We'll leave you with this quote from the 2018 results - they may be short on disclosure, but they're very long on technobabble:
"Kingdee created the unique KDDM Cloud dynamic domain model to support personalized Cloud service, and integrated social networking, artificial intelligence, Big Data and Blockchain technologies to provide scenario-based platform services for enterprise applications."