Long Alibaba – Plenty of Upside on This One

Alibaba logo

Long Alibaba @ $155/ADS (July 2017)

Target Price: $300+ (by CY2020 or before)

Summary: Recommend long BABA stock at $157, with 2 year price target of $250-300, further upside even after 2 years. 25% IRR from one of the world’s leading tech companies.

Introduction: Alibaba needs no introduction. The world’s largest e-commerce business by GMV ($580B USD), this company has revolutionary leadership in Jack Ma, who is insanely focused on the long-term. BABA’s ecosystem will get stronger over next two years as core retail, advertisements, video, cloud, payments, and investments all ramp up growth and deepen their moats. We forecast BABA to compound revenue, operating income, and FCF at 30% CAGR to 2021 and beyond, making BABA possibly a trillion dollar valuation in a few years, with stock price well over $300/share and market cap over $700B. Stock has run up recently (up 80% YTD) but is still far from fair value.

Core Assumptions:

  • China retail growth of 5-7% p.a. for next 6-8 years, driven by increase per capital income (mid-teens growth)
  • E-com retail penetration increase from 12% in 2016 to 33% in 2023, CAGR of 25-35%
  • # of T-mall users to grow in the high-single digits to mid-teens from 480 million in 2016 to over 800 million by 2024
    • Driven by increased penetration in Tier 1-3 cities and rural expansion
  • # of Taobao users to grow in the mid-single digits from 480 million in 2016 to ~600 million in 2024.
    • Lots of upside in rural areas
    • I don’t think accusations of Taobao selling “fake products” are accurate. Many Chinese prefer to purchase lower quality products knowing full well they could be fake, so clearly it’s not to deceive the customer, but because unique cultural differences indicate demand. Many US investors don’t really understand nuances of Chinese culture, but have slowly become more aware of these nuances.
  • T-Mall GMV growing at 20-25% p.a. for next 6 years at the very least (mobile contribution increasing to 85%)
    • Average spending per user increasing in the mid-teens (equivalent to 5-10% of annual income, which is itself growing at 10-15% per annum). No evidence BABA is reporting “fake” numbers, as short-sellers like to insinuate from time to time.
    • T-mall take rate growing from 2.6% in 2016 to possibly 4-5% by 2024 (with significant upside here if T-mall take rate approaches 8-15% seen in competitors, including AMZN, JD, Mercadolibre, etc….)
      • I see take-rate improvement, especially on mobile, being source of significant upside
    • Taobao GMV growing high-single digits:
      • Average spending per user increasing low-single digits
      • Taobao will see growth decelerate because of C2C shifting to B2C, but will still be key source of advertisement revenue and traffic source for T-Mall
    • My estimates are consistent with BABA reaching about USD 1 trillion in total US GMV by 2021, as Ma has said many times
    • Advertising revenue will be key source of upside, should increase at 25%+ rates for many years
      • China online advertising is $50B market, currently BABA has about 20% market share. Entire online advertising market will grow at 15-20% for next 8 years, I expect BABA to take incremental share and possibly increase to 30%+ market share of online ads in 5 years
      • BABA has massive user metrics that allow for personal, targeted advertising => increased ROI for advertisers. Similar to trends seen in the USA, Chinese ad budgets are moving very fast to digital
    • Cloud revenue should increase at very fast (triple digits) or high double-digits for many years, could increase from 4% of total revenues (2016) to over 20% by 2024.
      • AliCloud is #1 in China by a huge margin, we expect that to continue
      • LT operating margins could approach AWS standards (20-30%), we forecast 20% to be conservative
    • Revenue from international, cross border commerce should increase at 15-25% p.a. (at least) for several more years given BABA’s investments in promoting cross border consumption / trade / contribution from Lazada
    • Revenue from digital media / entertainment (Youku-Tudou, UCWeb, YunOS, AutoNavi, etc…) should increase at 10-15% clip for next 10 years, but won’t contribute much to overall revenues (only 2%)
    • BABA enjoys 50% EBITDA margins but margins could dip below 45% for many years due to LT investments (5-7 year time horizon). EBITDA (including and ex-SBC expense) should compound at 25%+ rates for next 5 years at least.
    • I think there is significant additional upside if BABA somehow finds a way to work with offline retailers in China to monetize a portion of the massive offline retail spend. BABA is currently working on unmanned supermarket concept stores similar to AMZN’s Go.

Equity Investments:

BABA investments

BABA has a ton of equity investments made over the years, summarized in the chart above. Core holdings include a 33% stake in Ant Financial, 30% stake in Weibo, and 47% in logistics network operator Cainiao.

Given it’s 50%+ market share in the fast growing mobile payments space, Ant Financial could be worth significantly more than $65B in a few years. Ant Fin currently has 50% market share in China’s rapidly growing 3P mobile payments space, and over 450 million annual active users and 150 million daily average users (much larger than the 200 million annual users for Paypal globally). Tenpay is #2 at 38% market share and should catch up a bit, but I believe the mobile payment space has enough runway for both BABA and Tencent to grow and take share. Ant Financial is currently the biggest contributor to value in BABA’s investment portfolio, worth currently about $8/share. I believe this will be conservative in a few years.

Alipay Tenpay share

BABA was also smart enough to invest in Didi rather than Uber a few years ago, the latter eventually retreating from China a-la Google fashion.

Based on latest fund-raising figures, the sum total of these investments equates to about $47 billion, or $18 USD / share.

BABA also has about $8B USD of net cash, or $3.1/share.

BABA’s capital allocation track record is phenomenal, as evidenced by this slide during BABA’s June 2017 investor day in Hangzhou. These guys really know how to create value and reinforce the ecosystem. Much better track record than JD.com, for instance.

The slide below says it all:

BABA investments track record.png

BABA has the best information and data insights into consumer spending in China (more commercial intent than Wechat), and will be able to effectively monetize this at 25% growth rates in the many years ahead.


  • 2573 million ADS outstanding.
  • Core biz (advertisements and retail) business should be valued on EV/EBITDA basis
    • My CY2020 EBITDA is RMB 285B, 13% higher than consensus. I expect BABA to surprise on take-rate and ad monetization upside
    • BABA will be compounding EBITDA at 20-30% p.a. for the next 5 years at least.
    • I used 18x EV/EBITDA on CY2020 EBITDA, minus net cash, minority interest / mezz equity => $750B equity value / 2,573M TSO = $290/share
    • Per share value of equity investments will probably increase from $18/share to $20+ by 2020. So $20/share (investments) + $290/share (core biz) = $310/share
    • Compared to $155/share today, assuming price is achieved in 3 years (market is forward looking so could achieve target before 3 years), yields 25% IRR at a minimum
    • If I’m a bit more aggressive, applying 25x EV/EBITDA on 2020 forecasts yields a $420/share valuation, or a 40% 3-year IRR. BABA would break a $1 trillion valuation, but it’s possible given the growth and market size.
  • Not surprisingly, given BABA’s vision, its execution, and size of China’s market, BABA could be one of the most valuable business on earth in 3-5 years.
  • Ma owns about 7% of BABA

Competition with JD.com:

The China e-com market is big enough for both BABA and JD to increase at comfortable rates over the next 5-10 years. BABA has around 80% of all China online GMV in 2016 (~98% in C2C and around 50% in B2C) and that will come down over time due to law of large numbers. But that doesn’t mean BABA can’t sustain 20-25% growth in GMV, with revenue upside from take-rate improvement and advertisements.

I don’t buy the argument that JD’s logistics service are definitely superior to BABA’s in any way. BABA has made significant improvements to its Cainiao logistics network and the historical service gap between Cainiao and JD’s own logistics is closing. Plus, a difference of 1 day in delivery doesn’t make or break the purchase decision for many consumers. Most Chinese still cite T-mall or Taobao as their preferred e-com site of choice in China.

BABA’s Ant Financial / Alipay will also make its retail, entertainment, and cloud platforms very sticky, another advantage over JD.com. The big data insights gleaned from having such a crucial pulse on payments will be very valuable for BABA over time.

Overall, I think there is room for both BABA and JD.com (and even VIPS) to grow over time. I think BABA’s focus on monetization of its user-base has huge upside and will cause the stock to re-rate over time.


There are lots of risks with investing in BABA, but I don’t believe any should keep an investor away from investing in one of the world’s greatest tech companies. Stock has run up recently (up 80% YTD) but is still far from fair value. Several risks also apply for all Chinese companies, so not much specific to BABA.

  1. Political risk
  2. RMB devaluation risk
  3. Integration risk
  4. Key man risk with Jack Ma
  5. Accounting risk – no hard evidence presented whatsoever by SEC or short-sellers
  6. Competitive risk – but I believe BABA will continue to deepen its moats and execute






VIP Shop – Is the Market Too Pessimistic?


Current Price: $11

Target Price: $16 (46% upside)

TSO (ADR) Diluted: 593 million

Market Cap: RMB 45B ($6.5B)

TEV: RMB 44.4B ($6.4B)

Free float: 29%

Daily Volume: $78 million USD

Founder (Eric Shen) Ownership: 15%


Long VIPS at current price of $11. Market is overly negative on VIPS despite near-term earnings misses. Nothing broken about the business model, management confident about 2017. Sentiment for Chinese ADRS has shifted from extreme optimism to extreme pessimism. VIPS down 66% from 2015 high of $30. Market tends to over-react with regard to Chinese ADRs on both ends, with extended periods of overvaluation and undervaluation. We are in the latter.

VIPS trades at 12x 2018 P/E on expected 20-30% EPS CAGR, and 7x rolling 2018 EV/EBITDA, a historical low. There is room in China e-commerce for multiple players and VIPS’s leading discount/flash-sales model, strong relationships with suppliers, and logistics capabilities should still have substantial TAM to grow into. Nobody knows what’s going to happen in 3-5 years, but any positive beat in 2017, buybacks, improving sentiment, and other catalysts can easily propel VIPS back into the $15-20/ADS territory, for a 50-100% return in less than a year.

There are rumors that VIPS is a takeout target given its net cash balance sheet and vertical category expertise, but rumors have been swirling for a long time so I won’t speculate on this.

Business Model:

VIPS was founded in Guangzhou in Dec 2008 by Eric Shen (Wenzhou native) and Xiaobo Hong (Vice Chairman). Eric was a born entrepreneur with a decade of experience in electronics products distribution before one day he saw his wife fixated on a French flash-sales site called Vente Privee in 2008. He realized that he would copy VP’s model to China where the market demand was huge. Having raised an initial 30 million RMB, VIPS set up shop in Guangzhou in 2008.

VIPS is a flash sales website selling discounted apparel, cosmetics, shoes, bags, home accessories, baby products, and other products via:http://www.vip.com/. It is basically an online version of TJX Maxx. China’s offline retail network is far more fragmented than the USA’s, thus giving e-commerce players like VIPS a big opportunity to rapidly expand coverage to entire China.

I won’t write a book about the company’s background as readers can find plenty of background information about VIPS online.

VIPS went public in March 2012, the worst time for Chinese IPOs in the USA given the reverse-merger scandals back then. Goldman priced the IPO at around $1/ADR but the offering was under-subscribed. Offering price ended up around $0.6, and stock traded as low as $0.4 at some point (all adjusted for 10 for 1 split in 2014). VIPS achieved full net profitability starting 2013 and the stock skyrocketed to a high of $30 in 2015.

2015 was a tough year. The huge China A-share bubble, short-seller attack on VIPS, and a few minor earnings misses killed VIPS. 2016 was no better, despite VIPS actually reporting a set of solid numbers. Market expectations were too high. The surprise election of Trump and fear of RMB currency depreciation continues to hurt sentiment about VIP and ADRS in general. Buzz about Alibaba and JD.com also took attention away from VIPS. VIPS fell 2/3 from its $30 high, trading at $11 today despite not-so-bad 2016 9M results (slight miss in 3Q16).


While VIPS might have been overvalued back in 2015, it is in value territory today.

Most investors don’t fully understand VIPS’s moat and culture. They are still creeped out by debunked short reports saying VIPS is a fraud. It’s not. Plenty of people in China shop on VIPS. Given fraud accusations have been going on since 2009, you would think Muddy Waters or some short seller would have dug out all skeletons by now. But so far, nothing. Eric Shen is extremely low profile, doesn’t do interviews, and doesn’t attend conferences. He is the anti-Jack Ma. Some say that Eric Shen is the only person in China that Jack Ma wants to meet so badly but cannot. Eric owns about 15% of VIPS stock.

Eric Shen is notoriously obsessed with customer service. When VIPS started in 2008, there were times when customers placed an order for a product only to find that product having sold out minutes before to another buyer. To honor these commitments, VIPS would actually buy another product at full price and ship to customer at 60%-80% discount, taking a loss to satisfy a customer’s orders. There are complaints of poor product quality from VIPS but that isn’t the same as VIPS selling fake products. The company is 100% focused on execution and growth. It does not try to manage the stock price. This discipline reminds me of Amazon’s religious devotion to customer service, although VIPS is not as innovative nor dominant as AMZN. But that’s ok, VIPS is cheap.

Active users in 3Q16 reached almost 45M, having grown at a 50% CAGR for the past 3-4 years. VIPS is a niche player, focusing on apparel and related categories only rather than everything (unlike JD.com). At 40-50M active users, it is much smaller than JD (200-300M active users) and BABA (>400M users).

The average customer orders 5 times a year, average order size RMB 200 ($29). These are not expensive items, but discount, fashionable apparel, shoes, bags, baby products, and cosmetics. Unlike BABA or JD’s 3P business, VIP stores products in its own warehouses on consignment and ships them to the customer directly. However, for items that it doesn’t sell, it can return them to the supplier.

Customers are mostly working class age females ranging from 20 to 50, and sales are spread evenly between Tier 1 & 2 (~50%) and lower tier cities (~50%). Buyers are part of the “working class” (工作族) in China and earn on average 60K RMB ($8K) per annum. With 5 orders of RMB 200/order, average customer spends about 1,000 RMB on VIPS, 2% of annual salary. This has upside.

The company has an army of 1,600 buyers whom it hires from other retailers, leading fashion brands, and fashion journals. It hires people who understands fashion sense and can shift through piles of unsold inventory at other retailers to identify products that will be easy to sell. Of the RMB 40B revenue in 2015, 39.4B (98%) was from VIP’s direct 1P model. The remaining 2% come from advertising fees, fees from 3P transactions, and fees from 3P brands using VIPS logistics).


The three largest categories are apparel (35% of rev), shoes/bags (14%), and cosmetics (13%) – total of 62% of revenues. Smaller categories include sports-wear, home accessories, and toys and baby products. Almost all of VIPS products cater to the female buyer, although new customers in recent quarters have been both male and female. 65% of revenue come from inventory clearance-related products, and the remaining 35% come from “in-season” or custom-made merchandise. In the past few years, VIPS has gradually increased the proportion of in-season and tailored merchandise as a result of it’s deeper relationships with brand partners.

Per 2016 consensus, China has 411 million females between 15 and 54 years of age – VIPS’s target audience.[1] I assume 80% of them access online shopping via mobile (82% of VIPS orders are via mobile) or PC. That’s 329M users. With estimated 50M users by 2016, VIPS has about 15% of the females in its core addressable age group.

User spending metrics have been pretty consistent since the company’s IPO. Growth is driven by increase in active members, as spending per order (200 RMB) and average number of orders/active member (5/year) remains fairly consistent. As of 3Q16, active members increased 43% yoy. Consistent with that trend, I expect total active customers for 2016 to increase 40% to 51M, and active members to CAGR at 23% to 2020. By 2020, I forecast VIPS to have about 102M customers, or about 27% of the addressable 15-54 female online shopper population by then.

VIPS is starting to acquire younger customers, post-90s, who tend to have slightly lower spending power. I take that into account by decreasing the average order size by single digit % for a few years out and keeping the outer years flat. I also run scenarios reducing order frequency.

On the competition front, JD is probably VIPS’s closest competitor. JD started its “flash sales” business in late 2014, so from the perspective of building relationships with upstream brands, optimizing purchasing decisions, and managing an extremely fast, unpredictable inventory turnover model, JD is lagging behind VIPS. However, JD does benefit from its enormous logistics network and overall large scale. However, JD is currently more focused on other areas such as fast-moving consumer goods (integrating the Yi Hao Dian acquisition), cold storage, managing its 3P platform, and expansion of logistics centers to rural places. I met with JD management in summer 2016 and it didn’t sound like flash sales was a huge priority for JD. It’s worth following, but in any case I think the market is large enough for both VIPS and JD to grow.

Overall, I forecast revenues to compound at 15-20% range (depending on assumptions) until at least 2020. Even then, VIPS will only have <1/3 of the total addressable female market (I make no assumptions about male customer capture), which I think is easily doable if the company executes as it has done historically.

Core part of being able to grow rapidly is having enough variety at the right price. VIPS generally manages to a blended gross profit margin of 23-25%, which has been quite stable over the past several years (and quarters). VIPS relies on its core team of buyers to build relationships with brand partners, retailers, and other inventory sources. These relationships are not easily replicable, hence VIPS’s moat. It is very similar to the moat at TJX or Dollar Tree – anyone else who wants to do flash sales at VIPS’s level has to hire a team of 1,600 of the brightest buyers and set up partnerships with over 8,500 brands (of which 1,600 sell exclusively to VIPS). The total number of brand partners grew from 1,075 in 2011 to over 10,000 at end of 3Q16, of which ~4,500 brand partners form a majority of the sales (remaining are thousands of small, independent brands). Revenue per brand partner has actually increased from just around $2M to ~5M in 2015 – meaning VIPS is increasing quality of partnership with each brand as well.


Based on TAM-analysis for key categories, I find that VIPS has huge growth potential and can grow at robust rates for at least another 8-10 years. See the chart below for my estimates of VIP’s penetration into each e-commerce product category. The highest penetration seems to be in cosmetics, home furnishing, and footwear, where VIPS has about 10% of the e-commerce market, whereas its core apparel category it is <5% penetration in e-commerce apparel, and mom & baby category is even less at about 2%. The key idea is that this data meshes well with VIPS management’s belief that VIPS has a single digit market share in key retail closeout categories in China and that the closeout business favors scale, making the business get “easier” as it gets larger. Many small brands want a single buyer to purchase all of its unsold inventory, and prefer a buyer like VIPS.

VIPS 3.png

About 5% of VIPS’s total GMV (same as retail sales for 1P) come from cross-border merchandise. VIPS started its cross-border business in late 2015 by partnering with overseas suppliers to bring coveted products into China via flash-sales model. Management has highlighted this areas as a growth driver going forward but it will probably be too small to move the needle for now. I assume the company purchases to a 23-25% gross profit margin for overseas sales as well. Management has guided that both inventory-clearance, in-season, and tailored products share similar gross profit margin of 23-25%. Margin fluctuations on a quarterly basis is mostly based on seasonal promotions (such as 11.11, 12.8 – VIPS founding date, summer clearances, etc…).

Gross margin for the remaining 2% of sales related to advertising, commission, and logistics service fees, I assume to be 70%. However, changing this assumptions doesn’t move the needle much.

On 2Q16 earnings call, management said “compared to the 400 million users in China, obviously we still have a very tiny percentage of that market share.” “To have more than 100 million active users, we’d probably need some time, but obviously, we think that would be the first milestone to hit … Eric is pretty confident, pretty optimistic in terms of where we can grow to in the future.” (~Millicent Tu). With estimated 50M active users for 2015, I think the 100 million user target is achievable. Achieving 100M users by 2020 (in 5 year) means 15-20% user growth per annum, and 100M users by 2020 is yet only 27% of my estimated population females ages 15-54 (371M, of which 85% shop online). Obviously this is an important metric to track but the numbers don’t sound unreasonable.

VIPS may even have a bit of counter-cyclical element. If the Chinese retail spending slows down tremendously that would mean more inventory for VIPS to help clear. Discount retailers tend to outperform during economic slowdowns, as can be seen from the stellar performances of USA closeout retailers from 2008 to 2013.

The four major expense buckets are fulfillment, marketing, tech, and G&A expenses.

Fulfillment: VIPS has followed JD’s footsteps in building its own warehouses and taking control of the logistics network. For most of its products, VIPS requires suppliers to store inventory in VIPS’s warehouses to ensure fast delivery upon sale (VIPS doesn’t take ownership of such inventory until the sale). VIPS’s flash sales model, coupled with Chinese consumer’s strong desire to get the product quickly, means that VIPS’s warehouse and delivery management has to be extremely efficient (customers want products within 2 days of placing order). By 3Q16, about 90% of VIPS’s orders were fulfilled by its in-house and invested last-mile delivery capabilities, with the remaining orders for 3rd party logistics companies. The company has logistics hubs in Guangdong, Jiangsu, Sichuan, Hubei provinces as well as in Tianjin. Additionally, company also leases bonded warehouses for growth of cross-border business.

Having owned warehouses decreases fulfillment expense as % of revenue but eats up capex. VIPS’s fulfillment expense has declined from 14% of sales in 2012 to 9% in 2015. Year-to-date 2016, fulfillment expense has continued to fall to 8.7% of sales. On a per-order basis, fulfillment costs have declined from about RMB 22 in 2013 to RMB 18 in 2015. Going forward, despite the company’s promises of reductions in rental costs, self-build warehouses, and increased efficiency in fulfillment (via automation), I keep fulfillment cost/order flat at 18 RMB/order and rental expense at 0.5% of sales, for conservatism. The lower order value (post-90s new users tend to spend a bit less) may also offset some of the increased in fulfillment efficiency. I expect fulfillment expense to be 8-9% of sales going forward.

For comparison, JD’s fulfillment cost as % of GMV is about 3-4%, about 1/3 of VIPS. That’s mainly because JD builds its own network of 200+ distribution centers across China and pays minimum rental expense. Moreover, because of its larger size, it is able to better amortize its fulfillment expenses (in a large part fixed) over a larger number of total orders. I don’t expect VIPS to reach JD’s fulfillment efficiency in the future due to these structural differences.



Marketing expense is the second largest cost bucket for VIPS. It has declined slightly over the past few years from 7% of VIPS revenue to about 5% currently. Recently in 2015, marketing expenses picked up slightly as VIPS is spending more advertising dollars in social media, network TV, and online video to capture the post-80s, 90s and millennial customers, who currently spend less, have lower stickiness, but who’s spending power is expected to go up with time. I personally have seen VIPS become more aggressive with advertising in several Chinese TV shows, including the 2016 blockbuster “Ode to Joy”. I don’t expect leverage at all in marketing expense as the company is aiming to reinvest all incremental profit into marketing to drive topline growth, as CFO Donghao Yang said on the 2Q16 earnings call. JD’s marketing expense is 2% of GMV but that’s because JD is much larger in scale and can leverage its expenses better than VIPS can.

I expect tech & content expense to remain 2.9-3% of sales going forward as VIPS will need constantly renovate its IT management systems. The company opened an R&D center in San Jose in 2014 to help analyze big data. JD’s tech & content expense is <1% of GMV because of its larger scale and better expense leverage.

G&A expense has been steady at around 3-4% of sales for VIPS, although I expect a slight uptick in 2016 due to VIPS’s rollout of its internet finance business (for customers and suppliers), although going forward this expense bucket should benefit from some leverage as well.

Capex has surged since 2014 since VIPS started building its own warehouses. Prior to 2014, capex was mainly used for leasehold improvements. Starting 2014, around 80% of capex went to building own fulfillment warehouses, capabilities, and infrastructure. 10% of capex is used to enhance IT systems and website, and 10% for other corporate purposes.

VIP’s owned warehouse space was 500K square meters in 2014, increased to 1 million sqm in 2015, and should increase to 2 million sqm in 2016. I understand that many short-seller reports (probably have been debunked already) cite VIPS’s capex and massive increase in warehouse space as specious, but many additional sell side reports came out later 2015 summarizing visits to the warehouses and found nothing out of the ordinary. I have not personally visited VIPS’s warehouses but I will trust that enough accusations and investigations on VIPS’s capex have been made that if something were really suspicious we would know about it already.

Capex was RMB 2.2 billion in 2015, about 6% of sales. I assume another 2.2B in capex for 2016 (in line with consensus) and similar levels of capex until 2019, then capex declining to 2% of sales for the outer years. JD spent RMB 5B+ in capex in 2015, so VIPS is spending about 40% of JD’s level. These numbers are consistent with VIPS’ goal of following in JD’s footsteps and building its own warehouses, upgrading current storage capacity, implementing more automation and sorting, and other changes.

Starting in 2016, VIPS started its own consumer “internet financing” business aimed at using its balance sheet to extend credit to customers. VIPS also extends credit to suppliers and has done so for 3 years. JD and BABA have had their own consumer financing units for years now so VIPS is trying to follow their footsteps. VIPS has said many times that the aim of the internet financing business is to help its core retail business. By extending credit to customers, VIPS can increase average revenue per customer by 30%. Via supplier financing, VIPS increases the quality of its partnerships with suppliers to capture more market share. Consumer financing is currently non-profitable but default rates are extremely low (between 0.2% and 0.5%). VIPS currently charges customers no interest for these short-term loans, in order to capture as many customers as it can. VIPS charges about 8-9% annual interest to suppliers. In the future, VIPS does plan to charge interest for consumer loans as well.

As of Sept 30, 2016, the total outstanding balance for consumer financing was RMB 1.47B, and RMB 525M for supplier financing. These numbers are still very small compared to the scale of VIPS’s overall operations, but VIPS has stepped up G&A spending to hire more staff to manage risk related to its financing business. Unlike JD, VIPS has no current plans to deconsolidate its financing business. The financing business does have a moderate impact on cash from operating activities but can be easily added back.

According to my “conservative estimates”, I expect VIPS revenue and EBITDA to compound at 22% CAGR from 2015 to 2020. Active customers will compound at 24% per annum over the same period to reach 107M by 2020 (30% of addressable online 15-54 aged females, 85% internet penetration rate). I expect gross profit margins to remain flat at around 24%, and minimum operating expense leverage. Based on my EBITDA and cash flow estimates, which are 20-30% lower than consensus EBITDA and cash flow projections for 2016-2019, VIPS trades at 10x 2018 rolling EV/EBITDA and 7.4x 2019 rolling EV/EBITDA, on a yet still strong 20% EBITDA growth profile. I expect core EBITDA margin to reach 6-7% in 2020 and possibly even higher beyond 2020 (TJX and ROST have mid-teens EBITDA margins). It’s actually NOT a good idea for VIPS to get its EBITDA margins too high because that would invite competition.

If you believe that VIPS can continue its strong operational performance and get slight improvement on operating and EBITDA margins by improving fulfillment, tech, marketing and G&A expense efficiency, then you get closer to consensus figures, which projects EBITDA compounding at 28% annually from 2015 to 2019. On those figures, VIPS is currently at 8x 2018 EBITDA and 7x 2019 EBITDA, a historical low.

Cash flow conversion from EBITDA is very strong due to the company’s negative working capital cycle – it receives payments from customers almost immediately and takes about 30-40 days to pay suppliers. This is similar to JD and all larger e-commerce platforms. Cash from operating activities has been 1-1.5x net income but heightened capex levels should eat up a portion of that from 2016-2019 as VIPS continues to build warehouse space. For simplicity, I give VIPS no benefit of the doubt and assume that self-owned warehouse capex is actually maintenance capex in order to protect its competitive position. Thus, I keep capex at elevated 2.5% of sales forever into perpetuity (we know 80% of capex seems to be growth capex, so this is conservative).

Net cash is 2% of market cap, and VIPS should continue to generate a high-teens ROIC (at least) given solid growth and margins. Given the stock price decline, VIPS spent RMB 650M ($100M USD) on share repurchases in late 2015, at average price of $16.14/ADS (46% higher than current price). That amounted to about 1% of the market cap, not too significant, but a positive signal nonetheless.

VIPS 5.png

Using a 10% WACC, RMB / USD at 6.9, and 2% terminal growth, my DCF yields $16/ADS price target, 46% upside. A sensitivity analysis is presented above. Nobody knows the future for sure, so maybe 3-5 year forecasts are useless. But my gut feeling is that expectations for VIPS have been reset from extremely high standards to very low standards. A reversion to the mean can take VIPS easily back to $20, although I don’t expect another $30 anytime soon. None of my projections give any value to the internet financing business because it is too small.

All it takes is 1 or 2 quarters of strong user growth, a good beat, or strong guidance to get the stock back to $15 or so. Despite competition from other e-commerce companies, I think VIPS is a strong player that really understands its niche. The balance sheet is solid and it generates enough cash for increased buybacks or deleveraging. At $6B market cap, it could be a takeout target by larger players like JD or Tencent as well, given general consolidation sentiment in Chinese internet space of late. I believe more good things can happen than bad at the current valuation, and the market isn’t pricing the positive scenario of strong user growth, per user spending, and margin improvements if VIPS continues to execute. The market is big enough for VIPS to grow much larger.

Key Risks:

  1. Competition intensifies
  2. China macro slowdown
  3. Privatisation (unlikely at this point)
  4. RMB depreciation
  5. Growth bottlenecks
  6. Margin deleveraging
  7. Capex overspend
  8. Trade wars

[1] http://www.indexmundi.com/china/demographics_profile.html

Turtle Beach Corp – Undervalued Microcap?

Sometimes you come across delicious opportunities in the small/micro cap space you can’t help but investigate…. When I first found this it was hovering at $1/share, but it ran up to $1.3 while I was writing this. I think the upside can be pretty good if video-game headset sales remain strong – $2/share is definitely possible with even more upside. The equity isn’t riskless given the debt, but the industry seems robust, the company is a leader, and the execution seems very shareholder-oriented.


Long Turtle Beach (NASDAQ: HEAR)

Price Target: $2.00+

Current Price: $1.00

Upside: 100%

Avg. Daily Vol: $360K (suitable for P/A or small funds)

Thesis: Long HEAR. Declining revenue from headset sales to older generation gaming consoles masking robust revenue and profitability growth in headset sales to new console cycle. HEAR is the undisputed market leader (40%+ dollar market share) in videogame audio headsets. 2016 marks the inflection point where revenue growth should recover. Additional accounting noise from Hyper-Sound business may soon dissipate as HEAR is exploring a potential sale. Sale of Hyper-Sound and removal of associated business expenses should reveal a much more robust core headset business. Rogue selling by former founder depressing stock but unrelated to fundamentals.

History: Turtle Beach, headquartered in San Diego, is a 41-year old producer of audio headsets. HEAR started producing video game headsets in 2005 and in 10 years became the market leader in this niche space, with top (40%+) market share in the US (its core market) and similarly strong market presence in Europe. Sales are about 72% US, 13% UK, 10% Europe, and 5% elsewhere (HEAR sells to over 40 countries in both offline and online channels).

In 2013, Turtle Beach merged with Parametric Sound, the original developer of what is now the Hyper-Sound segment of HEAR. Original Turtle Beach shareholders got 80% of the combined entity with 20% going to Parametric’s prior shareholders.[1] Thru the transaction, HEAR added Hyper-Sound has a new business segment. Hyper-Sound focuses on the development and commercialization of focused sound beams in home and outdoor settings, targeting people who have hearing issues (estimated 50M people in the USA alone – 1/7 of the population, have some degree of hearing loss[2]). Hyper-Sound is currently exploring opportunities to incorporate its sound-focus technologies in retail and commercial applications as well.

Turtle Beach has dominated the gaming headset market for the past decade. Sales are correlated with console cycles. Back in 2011-2012, Turtle Beach headsets had over 50% dollar market share of the US gaming headset market and did over $207 million in revenue in 2012. Competitors include Astro (owned by Skullcandy), Performance Design Products (PDP), Mad Catz (MCZ), and headsets produced by MSFT and SNY themselves. But Turtle Beach’s market share is far larger than the rest of the players in the market, due to design knowhow, excellent product quality, and strong marketing. Turtle Beach has around 40% market share in both US and UK. PDP (mid-teens market share), Sony (10% market share), and Microsoft (10% market share) are the next three largest competitors, so Turtle Beach is bigger than the next three competitors combined.

Sony and MSFT’s release of PS4 and Xbox One consoles in 2013 (the next generation) took analysts and HEAR by surprise. Sales of the prior generation products (PS3 and Xbox 360) rapidly declined as gamers upgraded. Unfortunately, sales of Turtle Beach headsets for PS3 and Xbox 360 declined as well starting in 2013. The company’s quickly designed and released products for the new generation consoles but the rapidly increasing sales of the new products could not completely offset the decline in headset sales for old generation consoles. Old headset revenue was a significant majority of revenue for HEAR in 2012 and 2013 but that proportion has come down to 20% in 2015 and an estimated 5% for 2016. HEAR stock traded as high at $21 back in 2013 but due to declining topline and margin contraction related to inventory clearances (lots of earnings misses) the stock fell all the way down to the $1 range where it sits now.

The continued investments in Hyper-Sound, an early-stage prototype concept with limited commercial application, added fuel to fire for HEAR. Hyper-Sound had zero revenue contribution in 2013 and in 2014 and 2015 revenue was a mere $1 million / year. Research and marketing expenses on Hyper-Sound was draining cash on HEAR’s headset business as the latter was concomitantly affected by unexpected announcement of the new PS and Xbox 4th generation.

Hyper-Sound hasn’t been fully commercialized yet but it is showing great promise as an innovative audio solution for focusing sound beams. Here is a short video that talks about the Hyper-Sound concept and the warm reception at E3 this year: https://www.youtube.com/watch?v=xbUahUG3Gwg.

The Opportunity:

For HEAR and investors, 2013-2015 was the “perfect storm” of sorts. The 4th generation console cycle came a little earlier than expected. Hyper-Sound continued to drain cash without commercialization breakthroughs. The company missed consensus expectations for several quarters in a row leading to its stock price falling like a rock.

Behind these unfortunate factors, Turtle Beach is actually enjoying strong growth and likewise leading market share in sales of gaming headsets to the new generation console customers. Management is confident Turtle Beach can maintain 40% market share over the next cycle. 2016 will be the last year of the “transition period” where sales of headsets designed for the 3rd generation should disappear, leaving 2017 and beyond figures to properly reflect the strong business momentum in headset sales for the 4th generation consoles.

Despite their mundane appearance, the gaming audio market (and audio in general) requires some tech knowhow. Traditionally strong players, including Turtle Beach, PDP, Skullcandy, Astro, and Mad Catz Interactive are getting only stronger. In 2011, Skullcandy acquired the gaming headset assets of Astro, further fueling market consolidation. The console makers like Sony or Microsoft sell their own headsets, but this is not a technological advantage or focus for them.

HEAR’s management believes that the 4th generation console cycle will not cause as much “disruption” to headset manufacturers as the sudden decline in the 3rd generation consoles did. Audio connectivity for headsets for the 4th generation are much more standardized and it appears that even if a 5th generation could come earlier than expected, 4th generation products will still be usable on 5th generation products.

Moreover, HEAR also believes that it will be able to take advantage of the virtual reality market. It already has introduced the Stealth 350VR headset for virtual reality games that enhances and amplifies the standard VR headset audio. The company believes that VR helmet manufacturers are mainly focused on the “visuals” and will work with audio companies like itself and Skullcandy on the audio side. It views VR as an opportunity (rather than a threat) to assert its expertise and dominance in video game audio solutions.

Moreover, despite the potential of Hyper-Sound, Turtle Beach announced on its 2Q16 earnings call back in Aug 2016 that it hired Piper Jaffray to explore “strategic options” for this segment, which includes a sale to someone else who can better monetize Hyper-Sound. HEAR hasn’t guided to any timeline for this but any sale of the technology would be a huge catalyst: (1) a sale would eliminate an expense drag on HEAR’s consolidated numbers to allow investors to appreciate the increasing profitability and cash flows of the core headset business, and (2) a sale would likely net a substantial amount of cash that HEAR can use to de-lever its balance sheet.

Why does this opportunity exist? For one, HEAR has a market-cap of only $48M which makes it off the radar for larger funds. Second, the tepid topline growth (which is caused by a declining sales to 3G consoles that will end in 2016) and lack of GAAP profitability means it is unlikely to show up on screeners. Third, Carmine Bonanno, one of the co-founders of Turtle Beach before the current CEO Juergen Stark was brought in by HEAR’s VC investor Stripes Capital in 2012 to run the company[3], once owned 10% of the company and has been selling without regard to valuation. Bonanno’s sales are probably: (1) to reduce his rather concentrated net worth exposure to HEAR, and (2) he does not like Stark’s combination of HEAR’s core headset business with Hyper-Sound, which he calls a “science experiment.”[4]

Carmine sells at regular intervals but has sold most of his original 5 million shares (10% stake). As of the latest Form 4 (filed September 21, 2016), he is down to 407.6K shares, so his selling is almost complete. Since I started writing this idea, the stock is already up 30% from $1.00 to $1.30 – part of this could be that Carmine is out of “powder” and can no longer depress the share price. Alternatively, if HEAR sells the Hyper-Sound business, Carmine might be appeased and stop selling when that happens. Whatever the reason, his selling is regular and appears more in protest rather than reflective of fundamentals. Given the $300-400K daily trading volume I would speculate that Carmine’s sales has been what’s preventing HEAR stock from going higher, despite strong 1Q16 and 2Q16 numbers. But that’s why the opportunity exists.


The best way to value HEAR is to consider its core headset business as a standalone business and guess as to how much Hyper-Sound could fetch in a sale as icing on the cake. It appears to me that HEAR is quite adamant about selling Hyper-Sound so this “sum-of-parts” valuation makes sense.

Going forward, Turtle Beach stock price will be driven by (1) the console cycle, (2) increase sales / market share of Turtle Bay products, and (3) profit margins and cash flow. The new console cycle appears to be at still early stages with strong sales in both consoles and game-related headsets. We are only a few years in the total life cycle of the PS4 and Xbox One and analysts predict that still 2/3 sales in this cycle are yet to come. 2015 marked the first time that the installed base of the fourth generation consoles exceed that of the third (see charts below taken from HEAR’s investor presentation).


Source: Turtle Bay May 2015 Investor Presentation

Industry participants expect the next few years to remain extremely strong for video game headsets – the market dominated by Turtle Beach with 40% market share. Below are several statements taken from public earnings calls of competitor Skullcandy (which sells gaming headsets under Astro brand) and Mad Catz Interactive. Newer version of the Xbox One and Play-Station 4, such as the Neo, will probably help the market grow as well.

  • “The gaming industry is an exciting industry, right, like everywhere you read, I think people are more excited about the growth opportunities in the industry. It’s – more people play games – and I was reading that the other day, more people play games now than they watch movies or listen to music” ~ Karen McGinnis, CEO of Mad Catz, 4Q16 earnings call
  • “Now to Astro, which is also killing it at retail… domestic sell-through increased 24% over last year ahead of strong teens industry growth… This occurred despite the fact that demand exceeded our supply during this past holiday season and retail shelves were literally bare of certain Astro’s SKUs at various times… We believe Astro can be a much bigger business… the gaming headphone market seems to be accelerating faster…Astro brand sales were exceptionally strong.” ~ Seth Darling, CEO of Skullcandy, 1Q16 earnings call
  • “I believe that Astro has the potential to grow into a dominant gaming company. Gaming is such a strong trend and Astro is at the top of the most desired performance products in gaming audio… This is a dynamic period for the gaming industry. Participating rates are growing rapidly as the new consoles continue to sell well.” ~ Seth Darling, CEO of Skullcandy, 1Q16 earnings call
  • “Gaming had such a strong end to year in Q4” ~ Jason Hodell, CFO and COO of Skullcandy, 4Q15 earnings call

HEAR ended 2015 with $163 million of revenue, 80% from headsets to new consoles and 20% from old-console headsets. Revenue has declined every year from the 2012 “peak” of $207 million, but that’s because of the rapid drop in old-console headsets. In the first half of 2016, revenue from new-console headsets increased 76%, but the continued decline of 65% in old-console revenue offset much of the consolidated growth, which beat expectations at 26% y-o-y. The real growth should come in 2017 and beyond as the drag from old-console headset sales disappears. I model 15% growth in each of the next three years as Turtle Bay continues to assert its dominant position in a mid-teens growth industry. HEAR has been able to take market share despite 40% market share already but I just assume industry level growth going forward. I expect HEAR to generate over $200 million in revenue by 2018 and possibly $230 million or higher in 2019, beating the previous “peak” back in 2012.

HEAR is guiding $12 million in adjusted EBITDA for 2016 from headset sales. That translates to about 7% EBITDA margins. EBITDA margin was 1.5% in 2015 due to the promotions related to clearing out old-console headset inventory. Back in the previous cycle peak, the company generated a 22% EBITDA margin on $207 million of revenue, so 7% margins for 2016 appears quite achievable. The company has mentioned that 10% EBITDA margins are probably more realistic going forward but I feel the experience of last few years had made them very conservative. Management believes 20% EBITDA margins (from 2011-2012) is not realistic – very few electronic hardware companies have 20% EBITDA margins. The company grew “too fast” back in 2011-2012 and didn’t invest enough, so they are targeting 10% margins going forward, and probably will move that a bit higher when it is reached. . At a 10% EBITDA margin, that’s $20 million of EBITDA, which compared to current TEV of $94 million, is a 4.7x EV/EBITDA.

The nearest comp is Skullcandy which has traded anywhere from 5-13x EBITDA over the past 5 years. However, SKUL includes a slower growing regular headset market in addition to their Astro gaming product line, which makes the comparison to HEAR’s pure, faster-growth gaming headset business imperfect. If anything, HEAR should be valued at the higher end of the 5-13x range given its dominance and growth.

As a crude reference, SKUL bought Astro in 2Q 2011 for $10.8M, or 1x sales. At 1x expected sales of ~$167M for 2016 (likely conservative), that translates into $167M of TEV, which translates into a $2.00-$2.30 stock price after netting out debt.[5] Takeout multiples are less precise but indicative that HEAR shares appear undervalued. A takeout could make sense at these valuations given that HEAR could be a strategic asset to a larger company like MSFT or Sony. Moreover, this assigns zero value for Hyper-Sound.

Two years ago, Wedbush published a report valuing Hyper-Sound at a $100 million valuation in itself.[6] The company also used a 12x EBITDA for Turtle Beach’s core headset business, but the numbers were caught off guard by rapid drop in revenue related to 3rd generation consoles. Hyper-Sound’s value today is anyone’s guess, but the technology seems to have improved in the past two years (management says that far more IPs have been filed since), despite uncertainties with commercialization.

The company wants to focus on the core headset business and has hired Piper Jaffrey to pursue strategic options with Hyper-Sound. The company is planning to either 1) pursue a sale altogether, 2) spin out Hyper-Sound as a separate publicly-traded vehicle, or 3) find a way to license Hyper-Sound to someone else for a royalty. Regardless of the result, management seems determined to monetize the inherent value of Hyper-Sound and focus on growing core headsets. My guess is they eventually find a buyer.

Just a few years ago, as a publicly listed company under Parametric Sound, Hyper-Sound had a market cap of over $130 million. Even if they sell for 1/3 of this value, that would yield $40+ million of cash that can be used to pay off the $54 million of debt and preferred liabilities – significantly deleveraging the company. As a quick exercise, if they sold Hyper-Sound for $40M of cash, net debt would come down to $14 million, TEV would be around $60 million, making the stock trade at 3x EV/EBITDA (assuming $20M EBITDA) for the core headset business – far too low.

HEAR also has $44 million of federal NOLs and $20 million of state NOLs that don’t expire until 2029. The significant NOLs could shield HEAR from tax expenses when pre-tax earnings turns positive. While this is more icing on the cake, note that the $64 million of NOLs > current market cap of $50 million, suggesting just how much HEAR equity may be undervalued if headset sales continue to do well for the foreseeable future. Management has said that the core headset business generates positive pre-tax profit and levered free cash flow – which means it will likely be able to realize the value of the NOLs once the Hyper-Sound related investments are removed from the financial statements to reveal “clean” headset segment figures.

While the company had $38 million of interest-bearing debt as June 30, 2016, none are due before 2018, and the company should generate enough EBITDA and cash flow to pay interest and principal upon due or to refinance the notes. In addition, while share issuances to fund debt repayments are also possible, we don’t think it is likely at current stock price level.

Insiders may be prohibited for purchasing more shares given HEAR is currently in strategic talks with Piper Jaffray about Hyper-Sound. Back in Feb 2016, the company did a private placement of $1.7M for $1.00 per share, and current CEO Juergen Stark purchased $150K at $1.00/share. He has not sold to date.

Stripes Capital LLC holds about 36% of the company via VTB Holdings. Moreover, the Chairman of HEAR’s board, Ronald Doornick (a partner at Stripes), holds another 3% stake. Juergen Stark, the current CEO, only holds 0.48% of the shares – we are a bit disappointed that his ownership level isn’t higher but we feel this doesn’t materially affect the thesis.

In July 2015, HEAR also issued to VTB and Ronald Doornick warrants to purchase 1.7 million shares of the HEAR common stock at $2.54 / share – which is over 100% higher than the current price. The warrants have a period of 5 years (expires July 22, 2020). Along with this, a variety of other small-sized warrant deals all have exercise prices far higher than the current stock price, so we believe management is correctly incentivized to get the stock price higher.   

tb chart 2.pngSource: Bloomberg

The chart below shows all insider buys (green) and sales (red) of HEAR stock in the past 3 years – pretty much all of the sales on the screen are price-indiscriminate sales by the original founders Fred (a little) and Carmine (a lot). Neither are involved in the current management of the company and both are retired. Carmine is actually almost done with selling his stake (as of Sept 22, 2016) but Fred could possibly start selling in the future which may temporarily depress stock price yet again.

tb chart 3.png

Source: Bloomberg 


Investing in microcaps is not without risks and HEAR is no exception. But I think the strong fundamental growth, HEAR’s dominance in video game headsets, the potential imminent sale of Hyper-Sound, and the termination of rogue, indiscriminant selling by Carmine, are all catalysts that could propel the stock much higher when the smoke clears. A takeout is also possible.


Nevertheless, possible concerns (non-exhaustive):

  • Competition or price wars in video-game headsets (HEAR’s core biz)
  • Reliance on retail channels
  • Obsolescence of technology
  • Surprises in the duration of the 4th generation console cycle
  • Hyper-Sound unable to fetch a buyer
  • F/X volatility and impact on overseas sales (core market is still US)
  • Mold or other product recall issues that harm sales/brand image
  • Inability to service $54 million of net debt (including preferred stock)
  • Manufacturing delays (Skullcandy alluded to this as a problem)
  • Active management (CEO, CFO) owning too little of stock
  • Additional equity issuance (mitigated if Hypersound is sold and cash used to de-risk the balance sheet)

[1] http://finance.yahoo.com/news/parametric-sound-corporation-merge-turtle-211000256.html

[2] HEAR 2015 10K, pg. 4

[3] See Stripes Capital’s page on Turtle Beach here: http://www.stripesgroup.com/investments/turtle-beach

[4] Call with HEAR Investor Relations

[5] [$167M TEV – 54M (net debt + preferred)] / 49M TSO = $2.30/share

[6] http://seekingalpha.com/news/1726323-wedbush-starts-parametric-at-outperform


TAL Education Group (XRS) – Expensive Chinese K-12 Education Provider but We Believe it Could Go Higher

TAL Education Group: Chinese K-12 Tutoring at an Expensive but Worthwhile Price

“As an educational company, if we do not teach our students well, that is no different from robbing and stealing from them.” ~ Zhang Bangxin (XRS founder, Chairman & CEO)


We recommend buying and holding TAL Education Group (NYSE: XRS). TAL is the market leader in traditional offline K-12 tutoring in China, has a long runway for growth, and is growing “owner earnings” at 20-30% p.a. Expensive valuation, but rock solid balance sheet (net cash), strong cash flow, smart management (37% stake), excellent track record, and optionality with online investments justifies current valuation. Chinese culture places strong priority on education – a cultural element foreigners often don’t appreciate. Stock could easily trade over $110 (75% upside) with 2 years.


Education is one of the most core elements of Chinese culture, dating back centuries since the foundation of the “Central Kingdom” itself. Career promotion and societal prestige were based primarily on examination scores since the Yang dynasty more than 1,500 years ago.[1] Since the “one-child policy” in 1979, education became even more important to parents as everyone wanted their kids to have a brighter future in China’s ultra-competitive economy. Education spending makes up an increasing proportion of most family’s expenditures – as high as 1/3 to ½ in some cities (and increasing in lower-tier cities and rural places).[2] Spending on after-school tutoring, extracurricular lessons, and other non-school education activities is on the rise and should continue to grow for many years to come.

Parents’ strong reluctance to take risks with their child’s education lies behind our long thesis for XRS. TAL Education Group is China’s leading K-12 extracurricular tutorial education company, offering comprehensive tutoring services to K-12 students covering core subjects, including math, Chinese, English, physical, chemistry, biology, history, geography, and political science. The company is better known in China as Xue Er Si (Chinese: 学而思 – “Learn yet Think”), after a well-known Confucius saying. In 2013, it officially changed its name to Hao Wei Lai (Chinese: 好未来 – “Bright Future”).

XRS provides tutoring services offered through 1) small classes done primarily under the “Peiyou” brand (85% of sales), 2) personalized premium one-to-one sessions under the “Zhikang” brand (11% of sales), and 3) online course offerings at xueersi.com and haibian.com (4% of sales). Small classes usually range from 15-35 students and typically meet on weekends or after-school hours on weekdays. XRS started investing in online offerings in 2010, but while online education is a huge emerging market in China, our research suggests that XRS’s core K-12 small class offering is difficult to duplicate online. XRS conducts its small class and personal tutor sessions across the years throughout 363 learning centers and 292 service centers in 25 cities in China. XRS started out in Beijing and its expansion in the past 5 years has been primarily in larger cities. We believe the white space in both existing markets and newer markets is tremendous.

Unlike its larger rival New Oriental (EDU), XRS’s bread-and-butter is the domestic Chinese K-12 curriculum, especially in mathematics, physics, and chemistry (math and science offerings make up 80% of its revenue). The brand is well-known and extremely well-trusted by Chinese parents, which makes it extremely sticky and de-risks expansion into new cities or towns. According to a brand survey by Brandz in 2015, XRS was ranked amongst the top 100 brands in China and the #2 education brand (only behind more recognized New Oriental).[3] EDU’s strength is in its English language curriculum and English-based tests like the SAT, GRE, or TOEFL.

EDU is also another interesting investment but we believe it’s focus on overseas test prep, while a growing market, isn’t as “safe” as XRS’s focus on the domestic Chinese K-12. If there was a sharp decline in volume of Chinese students seeking to study abroad, or a government legislation discouraging studying English or abroad, EDU would be affected far more than XRS. The chance of something negative affecting the development of China’s K-12 curriculum is far lower and that plays to XRS’s strengths.

TAL Education Group

XRS was founded in 2005 by Chinese by current Chairman and CEO Zhang Bangxin. XRS started with only a few learning centers which tutored students on Olympiad-level mathematics (a very popular course offering in China because those who excel in local or national math contests have far greater chance of admission to top universities). This is very different from EDU which started out as an English tutor. XRS is as local as you can get – it quickly expanded from covering just mathematics to physical, chemistry, and other subjects, including English later on. And while EDU tended to appeal more to high-schoolers and adults, XRS focused on tutoring kids starting from elementary school up to high school. XRS went public on NYSE in October 2010 at $10/ADS (2 ordinary shares per ADS). The stock is $63 today, representing a 5-year CAGR of 45%.

But at an only $5B market cap, we believe there is much more growth in store for XRS. XRS earned $1.12 per diluted EP-ADS in the year ending Feb 28, 2016, and we are aware it trades at a “scary” 55x trailing P/E. But high quality gems don’t come cheap and if you truly understand XRS’s growth potential and the industrial tailwinds, you will realize that the valuation can easily grow into rapidly-escalating earnings. Moreover, FCF/share averages 1.4x GAAP EPS and the stock trades at a more reasonable 30x 2018 P/FCF and 20x 2019 P/FCF for 30% long-term growth.

XRS founder

XRS founder Zhang Bangxin

We’ve translated Zhang’s 2016 address to XRS’s employees at the company’s HQ (attached at end) – we encourage you to read this carefully to get a sense for the founder’s long-term focus on winning not the next few years, but the next decade. We think not many investors appreciate Zhang’s longer-term vision. Note Zhang’s religious focus on service, delivering value to the customer, and self-innovation. His belief that “if XRS doesn’t teach students well that is equivalent to robbing or stealing from them” is setting the standard in education in China.

At Golden Harbor, we love to spend time appreciating the soft, cultural things that make a company tick and differentiate it from competitors. Our experience is that visionary leaders who focus deeply on customer satisfaction and having a clear mission tend to outperform over time and consistently do so. Zhang’s ambition, humility, and focus remind us of other visionary leaders like Jeff Bezos (Amazon), Jack Ma (Alibaba), or Richard Liu (JD.com) who, through their sheer determination and long-term thinking often not only dominate existing markets but create new ones.

The scope of Zhang’s ambitions for XRS over the next decade is extremely impressive to say the least – this is type of bold, focused leader we are comfortable investing along-side with for many years – even if the stock is optically expensive. And Zhang has his money where his mouth is – he owns 36.8% of the shares and has 74% voting power. Directors and officers own another 7.4%, making total insider ownership at a healthy 44% (88% of the voting power). We don’t mind the lack of voting power and want to be along for the ride. Zhang has mentioned in the press many times recently that he will not sacrifice high-speed growth for quality of growth. One of the XRS’s core values is “full customer satisfaction”. Zhang will not grow the company at a pace where thing start falling apart in any way.[4] So far, he has proven that 40% growth in enrollments and student count is a comfortable pace. We believe XRS will continue to deliver.

From 2005 to 2010, XRS was basically only in Beijing and Shanghai. The company didn’t start its China-wide expansion until the last five years (2011 – present), and we believe the growth is just getting started. In 2011, XRS operated 132 learning centers and 108 service centers in just six cities in China – with the majority of revenue (96%) still coming from Beijing and Shanghai. By 2016, XRS had grown to 363 learning centers and 292 service centers in 25 cities in China, with only 54% of revenue from the top 5 cities (Beijing, Shanghai, Guangzhou, Shenzhen and Tianjin). In 2015, XRS entered several second-tier cities such as Luoyang, Nanchang, Ningbo, Wuxi, and Fuzhou, and it entered Hefei (Anhui Province) in 2016. Over 5 years, the total number of learning and service centers increased by a 22% CAGR. The total number of students enrolled increased from 486,000 in 2011 to 2.31 million by end of 2015 – a 37% CAGR. Total revenue (reported in USD) increased from $111 million in 2011 to $620 million by FY2016 – a five-year CAGR of 41%. RMB-denominated revenue (which shows true growth) compounded slightly faster. Students enrolled per city has increased by about 17% p.a. for the past two years – suggesting that XRS’s expansion isn’t coming at the expense of quality on a per-city basis.

Geographically, XRS is just getting started. China has 62 cities with over 1 million people and 352 cities with populations between 100,000 and 1 million.[5] XRS is currently averaging 92,000 students per current city (only 25 so far). For cities with populations over 1 million, 92,000 students amounts to < 9.2% of the population. While this sounds high, keep in mind that many of these cities have far more than just 1 million.

The table below is illustrative. There are 62 cities in China with populations over 1 million. Together, the total population of all 62 cities is 262 million, or about 19% of China’s 1.4 billion population. Of those 62 cities, we’ve highlighted the 25 that XRS is currently in (in yellow) (note that XRS so far has only expanded to these >1 million population cities). We’ve made an educated guess that of the total 262 million people, 10% are K-12 age kids (say from 5 to 18 years old) that represents XRS’s target group. Personally we feel 10% is too conservative. We don’t factor in China’s recent elimination of the one-child policy which could give tailwinds to this estimate in next few years as babies grow into school-aged children. Even at 10%, that’s 26 million K-12 aged kids that XRS could tap into. XRS reported 2.3 million students at the end of FY 2016 – that’s only 8% of the addressable market. Even if you only take the cities that XRS is currently in, the total population is 165 million. Assuming again the 10% K-12 ratio, that’s 16.5 million K-12 in XRS’s existing markets. XRS’s latest 2.3 million students would only be 14% of this smaller addressable market.

XRS market analysis

Growth in revenue is driven by students enrolled. The company is guiding 40-50% annual growth in student enrollments for the many years to come. The population analysis allows us to sanity check this metric. XRS has only mid-teens market share in its existing markets – and that’s not even accounting for the 37 additional cities with greater than 1 million population that XRS has yet to expand in! And that doesn’t even include the remaining 80% of the Chinese population in over 730 cities with populations < 1 million! Given XRS is a leader in the K-12 space and kids can sign up for multiple classes (many take as many as 4 classes per year), that suggests that the white space for XRS is enormous. It could literally double the student count without expanding into any new cities and still double the revenue, EBIT, and net income of the business.

XRS has been net cash since its 2010 IPO. As of Feb 28, 2016, XRS had $247 million in net cash, or about 4% of the market cap. That’s puts TEV at $5.4 billion.[6] Dividing by the 2.3 students enrolled in FY2016 we get an EV/student of $2350. While that may seem high given the average student pays about $268 in tuition in 2016, we believe that figure will come down rapidly when XRS continues to add enrollment at the 30-40% clip for the next five years. At 35% uniform growth in enrollment, XRS will have 10 million students by 2021, and current TEV per student over 2021 estimate falls to $600, which is a bit more than 2x annually per student spending on tuition. New Oriental, with 3.6 million enrolled students and a $6.2 billion TEV, trades on a TEV/student of ~$1,709, but student enrollment is growing slower at 25% or so. While we’re not making judgments on whether this is reasonable valuation metric, our point is given that an average XRS student is far stickier than students of any other publicly-listed company, we feel there is room for the valuation to go up. Even at 10 million students, that’s not even 50% of the entire addressable market (> 26 million) in China’s 62 most-populous cities. And that doesn’t even include the 730 lower tier cities that XRS has yet to penetrate. Whatever the specific numbers are – we believe the market for XRS, the market leader with a market leading and well-trusted brand, is HUGE.

Keep in mind there is also room for ASP per student to go up. Currently, the average student spent $268 in FY2016 (ended Feb 28, 2016) for tuition at XRS courses. ASP for courses, on an apples-to-apples comparison, have increased about 5-7% per annum – in line with inflation. Overall ASP growth, at about 3-5%, b/c of mix shift (online courses, which are growing faster, have 50% lower prices than in class courses, and 1-1 tutoring is more expensive than small-class offerings). Management typically offers summer promotions (1 Yuan classes) to sign more students up (and students typically are sticky for several years) which depress pricing somewhat, but that is more than made-up when students elect several courses in the future at the full price.

As the chart below shows, the average family annual income in 2015 in large cities such as Shanghai was about $6,000 USD, whereas it was about 50% lower for more distant, western places like Gansu Province.[7] At $268/yr, the average tier 1 city family would spend 3-4% of annual income for XRS classes (per child), whereas families in less prosperous places would be spending about 7-10% of annual income (I do assume wages in lower-tier cities are growing faster than wages in first-tier cities). Education is an all-important investment for parents, and these figures, while high, are very reasonable in China. Parents also save for many years to pay for these services so they are definitely affordable, even though expensive.

XRS analysis 2

Although management has said that it can increase prices almost anytime (even in mature cities like Beijing and Shanghai where enrollment is growing mid-teens rather than 40%), Zhang prefers to play the long-game. Even without price increases, revenue can continue to grow at healthy 30-40% for many years. We don’t assume any price increases for our 8-year projection period.

Hiring and training great teachers is XRS’s biggest asset. Remuneration to teachers makes up its largest single cost bucket at ~50% of COGS. XRS hires graduates from China’s top universities (including Peking and Tsinghua Universities) and teachers undergo a 6 month training process before they are certified to lead classes themselves. Teaching expenses don’t appear to be scalable. From 2011 to 2016, the number of full time teachers increased from 1,169 to 6,594, a 41% CAGR. That lines up squarely with the 41% CAGR in revenue over the same period and shows that additional teachers will need to be hired to accommodate future increases in students (small classes disallow much cost leverage but increase student learning results because they get more 1-1 attention from teachers). We don’t mind this and actually think it’s a good thing that XRS hires more teachers to preserve the small-class feel that made its brand so trusted and well-known in China. We would actually be concerned if XRS skimped on hiring teachers and tried to conduct bigger classes. It turns out that the single digit ASP increases in student tuition are partially offset by single digit increases in teacher salaries, so we expect teaching costs to be neutral from a gross margin POV.

According to Jefferies research, the quality and experience of teacher is the single most important factor that parents consider when choosing education companies.[8] A natural question is whether or not XRS will run out of good teachers to hire – this may be an issue but neither the company nor its competitors have stated that top quality teachers are getting incrementally more difficult to hire. XRS actually tries to reduce its reliance on teacher quality per se by focusing on standardizing its teacher training programs so that it isn’t overly reliant on a few dominant teachers (who may leave and start their own start-ups!). XRS’s has an internal system to ensure that each teacher is trained up to rigorous standards and is well-paid and motivated. Teacher metrics are based on student success on examinations – very results driven. Teachers who perform well are promoted to larger cities and eventually corporate management positions. XRS doesn’t skimp on paying its most valuable employees and in return teachers provide students with above-and-beyond service. The teacher retention rate at XRS is over 90% – the highest in the industry.

XRS analysis 3

XRS rents nearly all of its classrooms in the 25 cities it is in. Interestingly enough, rent tends to go up yearly, and XRS’s total rental expenses CAGR’ed at 41% from 2011 to 2016 – exactly in line with revenue and teaching expenses. Given its “small-box” model, we expect no rental cost leverage and expect rental expense to hover at about 27% of total COGS. The company has not mentioned that space for expansion is an issue – but we will keep an eye on this.

The average learning center is about 750 square meters and this size has gone up over the past 5 years. We believe that XRS is opening up larger learning centers and hosting more classes in these centers to get operating leverage. Classes operate at only 20-25% utilization capacity because most courses are conducted on weekends.[9] We believe there is some room for capacity utilization (and margin) improvements but time will always be limited on the weekdays as students attend full-time school.

Other COGS includes small amount of stock-based compensation expense, course materials production cost, and depreciation of owned HQ building in Beijing. Given XRS is well-known for its high quality and proprietary course materials, we don’t think these expenses will come down as % of sales over time as they are fairly crucial to protecting XRS’s moat.

Since XRS’s 2011 IPO, gross profit margin has improved slightly from 49% in 2011 to 51% in 2016. We forecast future gross profit margin at 50% to factor in annual increases in teacher remuneration, lease costs, and more investment in developing better course materials. Gross profit grew from $54 million in 2011 to $316 million in 2016, a 42% per annum CAGR. We think continued expansion in student enrollments will continue to drive at least 30% p.a. gross profit margin growth going forward for the next 5 years at least.

While XRS says it relies on word-of-mouth as its sole advertising tool (which we feel is quite impressive), selling and marketing (S&M) expenses still make up ~12% of sales and has increased as a % of sales from 9% in 2011. XRS says this is due to increases in number of “sales and marketing personnel” and increased salaries per worker – in all honesty the company probably does a small amount of advertising – especially in new markets. Someone has to print the flyers and ads! In any case, we don’t expect any leverage from S&M expense going forward and keep it at 12% of sales (with room to come down).

G&A expenses went from 17% of revenue in 2011 to 26% in 2016. Most of this is due to XRS’s investment in content development and online education xueersi.com platform. Given XRS is in active investment mode in building out its online platforms (xueersi.com and haibian.com) and coming up with new courses and content (all proprietary), we expect this cost to remain elevated for the next 2 years at the very least (28% of sales) and gradually coming down as XRS reaps the results of its investments in the latter years (2019 and beyond).

Online Education – Peripheral, not Core

Over the past few years, there’s been a lot of discussion about China’s rapidly rising online education and its impact on traditional off-line operators like XRS and EDU. Pundits have offered spectacularly large market predictions for online education (some sources cite an RMB 200 billion market (30B USD) in 2015 and reaching almost RMB 700 billion (107B USD) by 2020) and spectacular growth rates exceeding 30% per annum.[10] Online education usually is in the form of pre-recorded lectures, or in recent years, live broadcasting by small number of teachers to larger number of students with interaction. Many tech companies are starting to invest in this field, including Baidu, Alibaba, and Tencent.

We think XRS is fairly well insulated from the online model (but we will be monitoring the ecosystem for potential disruptors). The first reason is that surveys show that parents are unwilling to consider online courses for core extra-curricular K-12 tutoring. Many students below the age of 18 lack discipline to “self-study” and given how important doing well in school and examinations are, parents won’t take the risk of allowing their kids to “slack off” with online-only programs.[11] Surveys show that while 44% of students have tried some type of online offering, 88% did not complete their online offerings, citing lack of motivation (36%) as the primarily reason and lack of proper learning atmosphere (aka classrooms and live teachers) as the second most important reason.[12] Online programs are usually more popular with adults who are more disciplined and know exactly what they want – it is not suited for students in the K-12 cohort. Secondly, core learning in K12 is mostly skill-based, and in-class environments are still the best for student-teacher and student-student interaction. Thirdly, the quality of many online-only players is often unproven and inconsistent. Online education start-ups don’t have the resources to hire and retain the best teachers.

Remember, it’s the parents that make most of the decision and surveys show that for them, the most important factor is instructor experience, and the second most important is learning environment. In fact, only 10% of parents (according to Jefferies survey) consider price to be the determining factor.[13] For many K-12 classes, the aim is partially to learn, but more importantly to pass the crucial examinations. Parents won’t skimp on testing out an unproven brand – they would still prefer established brand like XRS even if it’s more expensive. As disposable incomes rise in China, we believe the pricing power will work even more in XRS’s favor as consumers become even less price sensitive, but more brand-sensitive, in education spending.

Over the past two years, larger internet companies have tried to introduce educational offerings to broaden their categories. For example, Netease recently introduced a “cloud classroom” education offering tailored to professional learning for adults. For K12 Netease introduced their own “problem database” that users can access online. Baidu is focusing on introducing “problem databases” to allow students to download new problem sets. Tencent is taking a different approach and focusing on apps to allow students and teachers to organize education-related events (social networking is obviously Tencent’s core competency). Alibaba is introducing an “education platform” where teachers can sell services and problem sets, all replete with discounting and perhaps even haggling.[14] Even New Oriental’s Koolearn.com, the recipient of a $50M investment from an affiliate of Tencent back in Feb 2016, focuses on vocational training for adults.[15]

We are not worried about the recent IPO of $400M market cap online English-learning platform China Online Education (COE) either. COE, founded in 2011, is an online conversation-based English platform – if anything, we think COE’s model would be more of a threat to EDU than XRS. XRS derives over 80% of revenue from non-English subjects and isn’t dependent on English offerings as EDU or COE would be.[16]

We believe XRS will continue to growth amidst all this competition. Many of these competitors, while better capitalized, have corporate cultures better suited for their core specialties rather than education. Baidu’s core competence is search and advertising, Tencent’s is social networking, and Alibaba’s is commerce. Their education offerings seem too haphazard and distracted because it isn’t who they really are. Baidu is constantly thinking about how to make money off of advertisements on their education platform. Tencent is too focused on accumulating user volume on their education offerings. Alibaba seems more comfortable debating how to best monetize their “education supermarket” – commissions or advertisements? We prefer XRS (and even EDU) because “teaching students effectively” is hardwired into their DNA. They aren’t distracted, whereas education is at best a side-distraction for these larger competitors. They are unlikely to be putting their best people into developing an education offering, whereas XRS (and EDU) will be. Despite all of the investments that Baidu, Alibaba, and Tencent seem to making, none actually discuss education in their latest quarterly earnings calls. No mention at all! We also remember back in 2013 when YY.com was talking about making a big splash in online education thru their 100.com online education platform. Investors feared that this would be end of New Oriental and XRS.[17] Talk on 100.com lasted about 2 quarters and YY never mentioned it again. We think once this hullabaloo with online education calms down and XRS demonstrates consistent growth, the stock will go higher.

It’s not like XRS isn’t making its own moves in online education. XRS’s online strategy is to move slowly, leverage its core K-12 offline brands, and invest in other companies that are ancillary to XRS’s core strengths. XRS operates two core websites: xueersi.com and haibian.com. About 14% of its students, or about 300K, use these platforms. Xueersi.com focuses on pre-recorded video lessons and some live broadcast lessons, and haibian.com (the smaller of the two sites) focuses only on live broadcast lessons with interaction with the teachers regarding homework, problems, and other topics. Haibian.com has 20,000 students. XRS requires students using its online programs to respond within 15 minutes or otherwise it alerts parents. Although 14% of students at XRS use these online offerings, revenue from courses online makes up only 4% of XRS’s revenue – due to the lower pricing for online courses.[18] While XRS has said they are ramping up hiring of teachers in 2016 and 2017 (a margin drag) to beef up its online offerings (when developed, online should allow for some leverage because one teacher can broadcast to many students), we still feel that online will be at most complementary to the core off-line class setting. These online sites would be a great way for XRS to sign up students in cities it has not reached yet and get them to sign up for full-classes when XRS actually reaches those cities offline.

Besides its own online offerings, the table below shows XRS’s current online education portfolio. XRS has at least 10 more undisclosed investments beyond what is shown. We believe that even if online education were to take-off, brand still matters, and that the most reliable or trusted brand would be the most coveted online. We feel XRS is actually best positioned to take advantage of any upside should online education all of a sudden catch fire.

XRS analysis 4

Source: Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 79.

Financials & Valuation

We expect core operating income (EBIT) and EBITDA to compound in the 30%+ per annum range for the next 6 years at least, driven by strong execution, increase in enrollment, and penetration into new cities. At current $63 price tag and 4% of market cap in net cash, the stock trades at 47x forward EV/EBITDA for FY 2017 but for longer-term compounders we aren’t discouraged by this. The EV/EBITDA will quickly come down to 23x for FY2019, 17x for 2020, and 14x for 2021. By then, XRS won’t have even finished expanding in the top 62 cities in China yet and probably won’t even have started penetrated the remaining 80% of the population in lower-tier cities and countryside. Since its 2010 IPO, XRS has averaged 30x EV/EBITDA and 38x P/E over the past 6 years.

XRS analysis 5

One of the “wonderful” characteristics of XRS (as well as its peers) is that cash is received up front and expenses paid later. Expenses to teachers, staff usually take over 30 days to settle while cash is received several months in advance of when courses are offered. This item is usually booked as “deferred revenue” – a liability. By looking at changes in deferred revenue, we can actually estimate the growth in revenue 1-year out. XRS booked 289 million in deferred revenue by end of FY2016, a 63% increase over the 178 million booked at end of FY2015. If history is any guide, that means we could possibly be seeing 50-60%+ revenue growth for this coming year, well in excess of our 35% forecast.

XRS analysis 6

The negative working capital cycle allows XRS’s cash from operating activities to substantially exceed its net income. From 2010 to 2016, cash from operating activities was on average 2.12x net income. We model similarly strong cash flow conversion going forward. Capex has traditionally been 3-5% of sales (expect for 2012 when XRS spent $63 million purchasing a new Beijing headquarter). Capex is mainly used for purchase of computers, call center equipment, and other PP&E upgrades (remember, XRS leases all of its schools). We forecast capex as 5% of sales going forward and forecast forward FCF to average 1.4x of net income (XRS is unlevered). We expect FCF, or what some call “owner’s earnings”, to compound easily at 20-30% p.a. for the next 5 years.

For some who shudder at XRS’s 50+ P/E, let us say that the cash flow valuation looks a lot better, although it still might be very expensive for some. XRS did $145 million of trailing 12M FCF, which is a not-so-bad 2.7% yield on current TEV. Even at this valuation, with 2.7% yield and 20-30% growth in “owner earnings”, we believe buying XRS at $63 can work out if you have patience and a 3-5 year horizon. The stock went up 6x from 2011 to 2016 and we don’t see why the stock can’t go up 30-40% p.a. even at this valuation. We believe XRS can easily grow its student enrollment at 40%+ rates for many more years to come and that should drive the stock to go higher. Our FCF two years out yields 5% on current TEV and 3 years out yields 7%. While you may be able to get XRS at a cheaper price upon some type of sell-off, our experience tells us timing the market is utterly useless and over the long run it’s better to pay a bit more for a quality business than to pay less for an inferior-quality company. Based on our estimates, we wouldn’t bet against the stock reaching $100 / share within 2-3 years. A $100 price target corresponds to 25x our forecast of $4.00 FCF/share by FY2020.

Moreover, a TEV of $5B isn’t extraordinary if XRS successfully uses technology and its online investments to increase offline student stickiness. It is slightly smaller than EDU ($6.2B TEV) but if the recent New Oriental – Tencent collaboration could foreshadow further collaboration between tech companies and core education companies to come. It isn’t outside the realm of possibility to see XRS do a deal with one of the BAT[19] companies to fuse technology and education, or perhaps become a take-out target (given its strong cash flow, Fort-Knox balance sheet) altogether.

One interesting example of this “partnership optionality” is that between XRS and Didi Chuxing (China’s Uber) announced Jan 2016. XRS would provide up to $4.5 million (30 million RMB) worth of free education services and materials to the sons and daughters of Didi drivers as part of XRS’s public program to provide quality education to as many kids as possible. Over 90% of Didi drivers do not have time to tutor their children, especially on weekends when drivers are most busy and kids are not in school. Over 95% of Didi drivers say they would support these XRS gifts.[20]

It is interesting to note that although the shares have increased nearly six-fold since its 2010 IPO, Zhang has yet to sell a single one of his 59.55 million ordinary shares he held since the IPO (other officers have sold a little but insider ownership remains at 44.2% – down from 49% at the IPO).[21] This gives us extra conviction that Zhang feels bullish about the future. Zhang is only 35 – he has plenty of fight in him and plenty of time to see his vision play out.  

For those who are die-hard value investors, we would encourage waiting for XRS to pull back to $50 or below (which would be about 30x FY2017’s EPS of $1.50 and a 20% drop from today’s levels). Anything lower than that would be truly a gift – while we would welcome it with open arms, we don’t count on it.

Optically, the barrier to entry is low. Anyone can set up shop and offer tutoring services. According to iResearch, there are over 100,000 companies or institutions that provide extracurricular tutorial services in China.[22] But being able to replicate this model on a multi-city scale needs significant investment in content, quality teachers, and impeccable reputation – especially in K-12 segment. Zhang explained in a recent interview with the Chinese press that the education market is bigger, but completely unlike e-commerce or search, where scale allows duopolies or oligopolies to form (think of Alibaba – JD splitting B2C e-commerce, Ctrip and Qunar splitting OTA travel before they merged, or Baidu and Qihoo 360 splitting search).[23] In education, the addressable market is actually larger than a market like search, but it is composed of several smaller, vertical “niches” that cannot be tendered to by one big platform. There are many students with many needs, but you can only specialize in a few niches at once. There is core K-12 cohort. There is the English-learning and test-taking cohort. There is the professional education (for adults) cohort who want to learn specialized skills in IT or trade. It’s very hard for an education company to create a “platform” and take the whole market – the interaction with customer is not one-off as in buying a product, but over a period of months and even years. Teachers cannot be “jack of all trades” and must specialize.

We believe XRS’s strength lies in identifying this reality and focusing all its attention on optimizing its K12 product – its bread and butter. Zhang openly acknowledges that New Oriental is superior to XRS in foreign languages and that Tarena (TEDU) is more dominant in IT education for adults, but XRS knows where its strength lies. Every single initiative the company undergoes, from training its teachers to investing in other education companies, is done to strengthen this advantage. Fortunately, we believe K12 is also the largest market, has the stickiest customers, has strong pricing power, and offers the most room for XRS to expand. We do not find it surprising or unrealistic that XRS will, as Zhang delivers to his employees, be in over 100 cities over the next decade, from 25 today. The market is big enough and allows for scalability if you excel in your niche.

Parents won’t take risks with their child and send to potentially-inferior tutors just to save a dime (probably negative NPV investment from the parent’s view). Tutoring, and education, in China is about as delicate as infant formula – one wrong move and the child lags his peers and can’t catch up. Competition is seldom based on price. XRS is by far the nation’s most trusted K-12 brand and because of this we believe will continue to win market share. The stronger XRS becomes, the more it will be able to hire the best teachers (often stealing them from smaller, less well capitalized, localized tutoring shops). That in turn will lead more parents to pick XRS because it becomes the “low risk option”.

This virtuous cycle is further reinforced by the larger companies’ abilities to use technology to develop personalized data tracking for students. This will only increase user stickiness and increase switching costs. If a student who has been studying with XRS for a few years decides to switch, all of a sudden he/she loses all that data charting her progress for the past few years. Switching costs should only go up over the coming years.

New Oriental will probably do well over time too but we believe XRS’s model is more stable because K-12 is basically the backbone of China’s education system and essentially the nation’s future. That’s why we’re happy to compromise a bit on the valuation and ride the growth. Of course, at today’s valuation, there’s nothing wrong with buying a small stake and waiting for a pull-back to buy more.


Zhang Bangxin’s 2016 Address to Employees[24]

10-year Plan:

In the next 10 years, we will open offline schools in over 100 cities, with over 1,000 service centers and 10,000 classrooms. We will have over 100,000 plus full time employees. We will train over 1 million full time teachers. We will service over 20 million students offline alone, and combined with online, we will service over 100 million students. We will deliver over 1 billion lessons annually, which will exceed over 10 billion course hours. Our revenue will exceed 100 billion RMB (15 billion USD), and we will have over 10,000 recorded lessons on file.

Over the next 10 years, XRS won’t undergo just a simple change, but rather a fundamental transformation.

  • We will transform from a tutoring organization to an education organization. Over the past 12 years, XRS has accomplished many things, but we have focused on providing tutoring on weekends and in evenings on weekdays. But the future XRS will seek to penetrate deeper into the schooling system.
  • We will transform from an off-line business model to a technology-focused company.
  • We will transform from a Chinese company to a global company.
  • In the next 10 years, we will transform from an operations-driven company to a data-driven tech company

XRS’s strategic investments:

  • Toddler education – such as Baby-Tree, Xiao Ban Long, Heiha Technology, and Shark Park
  • Products in the K-12 space, including Xueke Net, Homework Box, and Qing Qing Home Tutoring
  • English products, including Shunshun Liu Xue, Li Bu English
  • Leading tech education companies overseas, including Minerva College, LTG, and game-oriented education products like Enuma

XRS has made many strategic investments over the years, but we won’t try to be everything at once. We hope to deepen and strengthen our core offerings, and leave the remaining to working with our investment partners to win-win.

3 Parables

The theme of my speech today is “Connecting You-Me in the Future”. I will first share 3 stories with everyone.

First off is a girl called Liu Zhixin. Liu grew up in Shandong province and came with her parents to Beijing during her first year of middle school, where she attended school near Renmin University. She also attended XRS’s Peiyou classes outside of school. She attended Peiyou for 6 years, where she took 3 subjects. She became the top scorer in 2015 college entrance examination in Beijing, and she was admitted to Peking University.

The second child is called Yuan Beibei. More than 10 years ago, when XRS only had tutoring classes in Beijing and only 2,400 students. That’s when Yuan came with her mom to sign up with us. She didn’t receive the top honors upon graduating high school, but she later was admitted to Berkeley University. She has not yet graduated, but she already has offers to work at Goldman Sachs and other prestigious organizations. Her mom followed XRS’s progress for many years and took a pay-cut to join us – becoming my first executive officer. Today, she is our VP Cai Yuli.

The third child is called Li Yige. She was a student of Beijing’s #4 Middle School. She didn’t end up going to Peking University, Tsinghua, Stanford, Harvard, but rather she was China’s first student to be attend Minerva. Minerva is a revolutionary type of college, founded by Harvard’s former principal, professors, and Stanford professors. The admission rate there is 2.9%. Many US students turn down other schools to attend Minerva.

Minerva uses the power of the internet to facilitate communication between teachers and students. Classes are conducted over 7 cities across the world through four years – the concept is very innovative. In 2014, XRS invested in this forward-facing university.

The reason I wanted to share with everyone these three stories is to show that the XRS today, compared to 10 years ago, isn’t just a tutoring organization helping students progress from elementary school to middle school and then to high school. Our students have demonstrated excellence not only in China but across the world.

We have focused resiliently on providing every child top quality education from 2003 to 2013. While our students have gone on to do great things, but we are always cautious. We must all remember these core three principles.

  • If we don’t teach students well, we are no different from stealing their money. 12 years ago, when we first started, we provided free trial classes for students and parents to test the courses out. We allowed refunds. This is not to say how good our business model is, but we started out on Day 1 by promising that we will not be a company that steals from customers. Parents are sending their kids to our organization, if we don’t teach them, we are essentially stealing their money. This philosophy sticks with me to this day and will continue to drive me going forward.
  • Secondly, we won’t earn respect if we don’t get students by word-of-mouth. China has many tutors, we have plenty of competition. Whenever we enter a new city or market, if we can’t provide students and parents something new and differentiated, why would we enter in the first place? We will focus on growing over the next few years, we are already in 19 cities, and we will enter several new ones this year.
  • Thirdly, a school that isn’t close with customers, doesn’t have a bright future. If we just focus on our business, profits, and not really understand the customer, spend time with them, understand their needs thoroughly, we will not have a bright future. We have worked on this for the past 12 years and want to continue improving in the future.

Let me share another story with everyone. Back in December 2013, when it was sign-up period at Shanghai’s winter XRS Peiyou classes, many parents waited in line until 11 – 12PM at night. At that time, one of our principals Yang Fuguang sent out a Wechat message to us showing a parent-packed picture – he said, “Parents are working so much to get their kids to sign up – we should have better action to show our appreciation for their trust.” I was thoroughly touched to see our products and services so welcomed by parents. Thinking back, this phenomenon also happened back in 2007 in Beijing. XRS sees over 90% signup rates for classes whenever it enters a new city – many people can’t even sign up. XRS still has many problems today, and many issues we need to fix. For one, we must find a way to provide our services and products to more cities to fulfill the great demand.

It’s going to be a long and arduous process. But we could at least consider ways to save parents the hassle of waiting in line till midnight. Last year, we introduced the XRS app, which has given students and parents more benefits.

We must create more competition for ourselves

Over these few years, what surprises me the most is, we are always striving to exceed ourselves. 3 quick examples – firstly, our Peiyou and Mobius classes. XRS Peiyou is the leader in K-12 education. Many of the teachers of our other brands come from their experience teaching in Peiyou. Over the past 7 years, we have continued to ponder, whether there will be a product that will exceed Peiyou in some or all respects?

Over the last five years, we have often wondered whether xueersi.com (our online platform) will exceed Peiyou. Will we even need off-line education? We eventually realized, that online education and offline education weren’t the same thing. Then, what will off-line education look like in 5 years?  We created Mobby.com, which uses iPads to teach students, using many new technology. Later on, under competitive pressure, XRS’s Peiyou wasn’t just not defeated, but it grew even stronger. Our work with Mobby allowed us to gain better perspective, to understand how to connect our product with technology better. We are always trying to give exceed and reinvent ourselves.

Another example is Le Jia Le and Li Bu. XRS started as focusing on the sciences, but 7-8 years earlier we entered the English market for young kids. Le Jia Le is becoming a national level English brand under the Peiyou umbrella. Last year, we acquired Li Bu English – which is focused on students who want to study abroad or attend international school. Along with our other English product offerings – Le Wai Jiao and Shun Shun, we have a complete English product offering.

The third example is our xueersi.com online school and Haibian direct broadcast offering. Our online school is a core pillar of our brand and growth strategy. Seven years ago, when we started our xueersi.com online offering, we looked up to our Korean peer Megastudy, which did very well in Korea but was overtaken by us in China. Xueersi.com was the first in China to use high-definition broadcasting of lectures and later on introduced direct broadcast in addition to recorded lessons. Over the past few years, our combination of recorded lectures and new innovations in direct broadcasting has become a global standard and first.

Even in spite of this, over the past three years, we also incubated yet another online product – Haibian (Chinese – “Seaside”) direct broadcasting. Haibian uses direct broadcasting plus service to help parents and earn good results and reputation. Xueersi.com and Haibian.com compete but also complement each other. They learn from each other and compete with each other, taking each other’s technologies and lessons learned.

Past several years, in addition to perfecting our core K12 product, we have used M&A to expand our horizons. We have invested in many education technology companies, most are #1 in their respective fields. For example, Baby Tree is #1 in the maternity category.

In sum, XRS will continue to expand its assortments with K-12 as the core. We will enter baby and young child market via Baby Tree, Xiao Long Ban, Hei Ha Tech, and Shark Park. We will expand in our core K12 offering, with brands like Xue Keji, Homework Box, Qing Qing Tutor. We will also expand in English and foreign language offerings with brands like Shun Shun, and Li Bu English. We will invest in the leaders of education abroad, such as Minerva, LTG and Enuma.

Data Technology

Education is about to welcome the data technology age. What do I mean by this? Everyone is aware that big data has ushered in rapid developments in society. The future of education sector will be predicated on big data. We have realized that the internet and information age are transforming the education sector, but everything revolve around service. In the age of education 1.0, everything revolved around teachers. In 2.0, everything revolved around the student, with a focus on motivating and encouraging students.

The education industry of today has its roots in service. Will big data change this? Actually, when service meets content, service is the core, the content is the ancillary. When service meets tools, service is still the core, tools are the ancillary. When service meets direct broadcast, service is still the core, broadcast the peripheral. But in the next decade, besides service, there will be big data. Data will complement service as the core of our business.

Why is data so important? Data helps us make education more individualized. Current education treats humans like machines – regardless of a student’s capabilities or interests, we feed him/her the same thing. This is a very low-efficiency way of doing things. In the world driven by data, a 20-year old of the future can get a Master’s Degree – won’t need to wait until she is 28 or 30. Hence, data and analytics will underpin the direction of the education sector for the next decade. We have just invested in a global-leading self-adaptive learning platform called Knewton. Knewton has one of the world’s most comprehensive databases – we’ve also signed service contracts with them.

In the past, teachers and tutors had always been paid by the hour. To some degree, we also paid by time taught today, similar to barbers and nail artists. In the future, education will be measured by effectiveness and results. Those organizations who can allow students to improve the most will be the winners.

Looking even future, we may even measure success in education based on unleashing the student’s potential and longer-term growth. We may be able to earn our tuition only if we help a student achieve success in society, maybe if they become an expert in some industry or practice. That will be the true value of education and how we should progress.


[1] http://www.chinahighlights.com/travelguide/ancient-education.htm

[2] http://www.eastasiaforum.org/2014/03/15/education-chinas-most-important-economic-weapon/

[3] https://www.millwardbrown.com/docs/default-source/global-brandz-downloads/china/BrandZ_2015_China_Top100_Chart_EN.pdf

[4] http://www.duozhi.com/leader/201501272754.shtml

[5] http://worldpopulationreview.com/countries/china-population/major-cities-in-china/

[6] Note that XRS’s basic ADS outstanding is 80M and diluted ADS outstanding is 91.5M. We use diluted metrics to get more “conservative” calculations.

[7] Extrapolated from 2012 survey numbers as reported at: http://www.nytimes.com/2013/07/20/world/asia/survey-in-china-shows-wide-income-gap.html

[8] Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 9.

[9] JP Morgan report on XRS (Leon Chik & Christine Wang): “Lifting PT to $70 – Tech savvy for better growth.” 23 June 2016.

[10] Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 13.

[11] According to a Sohu survey in Nov 2015, where some 200,000 respondents participated, results show that while 83% of students participate in extracurricular courses, 68% of parents say they would not consider online tutorial courses for children – see Jefferies report pg. 16.

[12] Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 19.

[13] Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 17-18.

[14] http://www.huxiu.com/article/136930/1.html

[15] http://www.prnewswire.com/news-releases/new-oriental-announces-tencents-investment-in-xun-cheng-and-xun-chengs-proposed-ipo-in-china-300212652.html

[16] JP Morgan report on XRS (Leon Chik & Christine Wang): “Lifting PT to $70 – Tech savvy for better growth.” 23 June 2016.

[17] http://technode.com/2014/02/25/yy-launches-a-separate-online-education-platform-100/

[18] Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 37.

[19] BAT stands for China’s three largest technology / e-commerce companies: Baidu, Alibaba, and Tencent.

[20] http://edu.sina.com.cn/zxx/2016-01-13/doc-ifxnkkux1264295.shtml

[21] See “Share Ownership” in XRS 20-F.

[22] Jefferies report by Johnny Wong and Kevin Zhao: “Education: Powering Ahead: Assuming Coverage of Chinese Education Sector.” 8 July 2016. Pg. 9.

[23] http://www.duozhi.com/leader/201501272754.shtml

[24] http://tech.sina.com.cn/i/2016-01-20/doc-ifxnuvxh5029913.shtml