Trade War as Prisoner’s Dilemma

A trade war is like a Prisoner’s Dilemma game gone wrong.

  • (1) Trust between both sides has eroded.
  • (2) They cannot communicate effectively.
  • (3) They decide to forego cooperation and defect.
  • (4) And, by foregoing cooperation both sides are hurt.

In a prisoner’s dilemma game, two players are separated (can’t communicate) and given two choices: cooperate or defect. The only winning outcome is when both players choose to cooperate. 

There is a special kind of prisoner’s dilemma game, where the game is repeated many times over and the players can remember the other player’s actions. This means past rounds matter, reputation matters. In these prisoner’s dilemma games, a winning strategy emerges: Tit-for-Tat.

How Tit-for-Tat wins Prisoner’s Dilemma games

First off, the Tit-for-Tat strategies that win are slightly modified. They have a function that tries to break a negative ‘defection’ cycle by forgiving the other party and making the first move to cooperate. 

Forget injuries, never forget kindness

More specifically the winning strategies tend to be: “nice (never the first to defect), retaliating (willing to defect), forgiving (willing to attempt to regain trust by breaking a defection cycle), and non-envious (not specifically attempt to outscore individual opponents)”[1]. I’m going to focus on forgiving because that is where the cycle gets broken.

In a standard Tit-for-Tat strategy you copy the move of your opponent. If they defect : you defect, if they cooperate : you cooperate. This easily turns into a negative cycle, where both sides keep defecting (and spending a lot of extra time in jail).

The negative cycle is broken first by forgiveness and second by extending a hand (attempting cooperation). One side decides to forgive and tries cooperation for two rounds.

Why two rounds? Because the players don’t know the other player’s action until after the round. The first round forgiveness is the signal. The second round is the real chance for cooperation:

  • If the other side cooperates, the negative cycle is broken. (Hurrah!)
  • If the other side refuses cooperation, in the next round (the third) the forgiving player switches back to escalation. The negative cycle continues.

Here’s some food for thought on forgiveness:

  • “记人之善,忘人之过” – 三国志 forget injuries, never forget kindness.
  • “We must develop and maintain the capacity to forgive. He who is devoid of the power to forgive is devoid of the power to love. There is some good in the worst of us and some evil in the best of us. When we discover this, we are less prone to hate our enemies.” – Martin Luther King, Jr.
  • “To forgive is to set a prisoner free and discover that the prisoner was you.” – Lewis B. Smedes

Forgiveness is the intentional and voluntary process by which a victim undergoes a change in feelings and attitude regarding an offense, lets go of negative emotions such as vengefulness, forswears recompense from or punishment of the offender, however legally or morally justified it might be, and with an increased ability to wish the offender well.

Wikipedia [2]

Thanks for reading. If you want to be notified of new posts, please subscribe. I welcome your thoughts and comments. Continue the conversation in the comments, on Twitter @jameshullx or on LinkedIn.

Assets: Atoms versus Electrons

In this post I focus on the assets of a business, stuff that is owned by the business and is fundamental to the business model. Human capital is fundamental as well, but I will leave that for another day.

Note: the impetus of this post was a conversation with Elliott and Michael on China Tech Investor podcast (23): Is Luckin Coffee a real business? I believe Luckin Coffee is using scaling techniques that work for electron businesses but are not suited for atom businesses.

A post on Luckin Coffee where I will share my analysis on their traffic is in the works, you can follow this blog to receive posts by email. — And on to the post.

There is a difference between atom businesses (primarily focused on physical) and electron businesses (primarily focused on digital). Hat tip to Josh Wolfe for this concept.

Electron businesses are primarily media and software (in some cases intellectual property). They are supported by highly efficient infrastructure: fiber optic and copper data transfer, electricity generation and distribution, and electronic devices. Media and software is codified and stored on physical devices (sometimes in electrons), and can be infinitely flawlessly replicated. Storage costs for media and software are near zero: a hard drive on a laptop. Operating cost for software is low and can be “on demand”. Media operating cost is more about keeping it organized and available. Media and software have production cost and distribution cost. After media is produced it is in storage awaiting distribution. Maintenance costs for media are near zero. For software there are maintenance costs, they can range from high to low depending on complexity of the software and the rate of change in the underlying physical layer.

Atom businesses are primarily composed of physical things: assets or —stuff. All assets degrade over time to the point they are useless. Proper care and maintenance can extend an asset’s life, but it’s not free. Physical businesses also require “space” in the form of land or a physical location, which have to be found and acquired and come with carrying costs (such as leases, taxes). Physical locations (land, stores) are not easily spun-up or spun-down, unless a contractual-layer in the middle makes it so (such as a business renting month-to-month from WeWork).

Contractual-layer businesses redefine the contractual relationship on either the supply-side or the demand-side, usually the supply-side. For example, Uber redefined the supply-side for hired drivers (taxis, private car services) by allowing anyone with a phone and a car to become a driver, in effect circumventing the hiring/vetting process or stringent cab-driver license requirements. This lowered the barrier for drivers on the supply side. On the demand side, Uber used infrastructure available (electronic devices and GPS) to replace location-based hailing (standing on the sidewalk) and phone-call hailing with an app on a device with GPS.

Thanks for reading. I welcome your thoughts and comments. Tell me how I’m wrong on Twitter @jameshullx or on LinkedIn.

CTI20 Notes – Xiaomi 4Q18 earnings

Disclaimer: These notes are shared for informational and educational purposes only. THIS IS NOT INVESTMENT ADVICE.

These notes were composed on April 7, 2019 for the China Tech Investor podcast episode 20, listen on Xiaomi is traded under 1810 in Hong Kong and ADRs under XIACY.

Revenue growth

Source: Xiaomi Q4’2018 investor presentation
** IoT and lifestyle includes laptops and TVs

International revenue reached RMB 70 billion, YoY growth of 118.1%, about 40% of total revenues.

Xiaomi (and Huawei) is growing in an industry that is shrinking. That means they are gaining market share.

Source: Xiaomi investor presentation, IDC

They have increased their AI patent applications in the 3 year period ’16-’18 significantly from ’13-’15.

Source: Xiaomi investor presentation,

Some new products launched in 2018, IoT and Lifestyle segment:

  • Air conditioner (July 2018)
  • Washing machine (Dec 2018)

Key IoT products, excluding TVs and laptops:

  • Wearable watch
  • Electric scooter
  • Floor sweeping robot

IoT platform has 151 million connected devices as of Dec 2018, YoY growth of 193.2%

Teaming up with IKEA

All IKEA smart lighting products can be connected to Xiaomi IoT platform and controlled by Xiaomi AI Assistant and Mi Home App

Interesting note: 50%+ Mi Home App users are non-xiaomi phone users

Internet Services

Source: Xiaomi investor presentation

International Market Share

  • #1 in India for six quarters, Canalys estimates just south of 30% market share
  • #2 in Indonesia, 21% market share, after Samsung with 24.5% market share
  • By shipment growth, #4 in Western Europe. BUT market share still low (3.2%) compared to Samsung (32%), Apple (26%) and Huawei (20%)


Lei Jun’s April 25, 2018 blog post: “At the Mi 6X launch event in Wuhan earlier today, we promised our users that our hardware business, including smartphones, IoT and lifestyle products, will have an overall net profit margin that will never exceed 5%.”


And the same pledge from Xiaomi Chairman’s letter to shareholders:


Some financials here:

Source: company filings, Hullx calculations

Inventories saw a spike upward. An increase in inventories, especially finished goods could mean three things: (1) products are not selling as fast as in prior periods, (2) management decided to boost inventories for expected future sales, or (3) a bit of both.

Management addressed this on the call and in their annual filing. It seems they are trying to “get ahead” of the negative implications of increasing inventories.

“…an increase in inventories of RMB4.9 billion. The increase in inventories was due to the decline in sales of our smartphones in the fourth quarter of 2018 and our procurement in preparation for the new product launches. Our inventories as of January 31, 2019 declined 12% comparing to that as of December 31, 2018, according to management accounts, primarily attributed to the increase in sales of our smartphones post the launch of new smartphone models. Our inventory turnover days was 55 days in January 2019, excluding the inventories related to real estate business, according to management accounts.”

Free Cash Flow:

  • Cash used in operations: RMB 6.2 billion
  • Less Capex of RMB 2.3 billion
  • Gives, free cash flow of Negative RMB 8.5 billion
  • Cash and Cash equivalents was RMB 30.2 billion, ST bank deposits were RMB 1.3 billion

Earnings Call notes

Lei Jun started off talking about revenue growth, IDC rankings, design awards, quality improvements…

Multi-brand strategy: Redmi and Xiaomi

  • Redmi will pursue the highest price-performance ratio with the best quality in the market.
  • Xiaomi will be used to penetrate the mid to high-end market.

— Are they that different?

Lei Jun spent some time talking about demand for Xiaomi 9 outstripping demand. He addressed a rumor that the company is purposefully withholding supply, announcing Xiaomi 9 supply is likely to exceed 1.5 million units by end of this month (March). — Comment: Kind of odd addressing customers on a quarterly earnings call.

Question: ARPU trend. Q4 sequential quarter decline in advertising revenue and ARPU

“Revenue from our internet services segment decreased by 14.6% from RMB4.7 billion for the third quarter of 2018 to RMB4.0 billion for the fourth quarter of 2018, primarily due to decreased advertising revenue.” (company filing excerpt)

Answer: (1) international internet services is lower, but international has added MAU (the denominator in ARPU) and (2) China domestic advertising market, gaming in particular affected them.

Question: 2018 smartphone gross margins took a dip. Should we expect margins to revert back to 2017 or continue in line with 2018?

Answer: Because we promised we will make no more than 5% income margin on hardware, before we went public, our goal internal actually is to make 2% to 3% in the future, so the number that we focus on is hardware net income margin.

Lei Jun: 我们的目标就是两三个点所以我不是特别关心毛利。My translation: Our goal is 2% to 3% so that’s why I don’t care about the gross profit.

Lei Jun (CEO and Founder) has three main focuses:

  1. Net promoter score, a reflection of customer satisfaction
  2. Smartphone shipments
  3. Deepen internet services offering

CTI19 Notes – Tencent 4Q18 Earnings

Disclaimer: These notes are shared for informational and educational purposes only. THIS IS NOT INVESTMENT ADVICE.

These notes were composed on March 31, 2019 for the China Tech Investor podcast episode 19, listen on Tencent is traded under 0700 in Hong Kong and ADRs under TCEHY.

Here is Tencent’s Q4 2018 investor presentation

Mini Programs

  • working well, but need developers. Working with over 100 universities to include MPs in their curriculums. Students make up 24% of MP developers in China


  • Revenues doubled to RMB 9.1 B
  • Paying customers more than doubled for Q4
  • “Clear market leader for online games and video customers”

Retail → Digital transformation

  • customer engagement with CRM & data analytics
  • lot of buzzwords: marketing ROI, customer targeting, AI, LBS, and Big Data, Smart Solutions

Strategic Upgrade:

Shek Hon Lo (CFO & Senior VP): “I would like to let you know that following our strategy — strategic upgrade from the consumer Internet to the industrial Internet, we will add a new revenue segment to better reflect our evolving business mix in this quarter. “

What is the new segment going to be?
My two guesses: financial or cloud services.

Earnings Call Q&A notes:

Online video:

  • Delay in online video content launches
  • “costume drama series” relatively important to drive advertising and subscription revenue.

Games pipeline:

  • 7 new smartphone approved, 1 PC game, since resumption of BanHao
  • several dozen more in pipeline
  • Backlog of approval will likely have some impact on industry growth

Advertising margin:

  • High margin: weixin moments
  • Middle: Official Accounts
  • Low: video entertainment content

2019 Gross Margin, puts and takes:

  • WeChat pay:
    1. Take revenue from merchants and customers
    2. Big chunk goes to bank charges
    3. No interest income from reserve cash
    4. Trying to compensate with fintech products
    5. – *IF* bank charges get reformed, it’d be a big upside for Tencent
  • Pony Ma showing some real integrity here, pouring some cool water on WeChat’s payment platform. After talking about Gross Margins:
    • “But below the gross margin, there’s actually a marketing cost, which is actually also very high, because despite the fact that we have positive gross margin on the growth side, we are actually engaged in a lot of marketing activities, including consumer subsidies, including some merchant rebates in order for us to continue to build market share and also make our payment mechanism — I mean our payment platform more popular with merchants.”
    • Tencent will run payment platform as an infrastructure service, meaning lower margins
  • Content costs rising.
    • Should be recurring. Their business is a content business. Games is content. Video is content. Music is content.
    • Tencent already is a digital services business.

Free Cash Flow is growing

  • Q4’18 up 18% yoy to RMB 28.6 B
  • FY18 up 11% yoy to RMB 83.4 B

CTI18 Notes – Meituan 4Q18 – Mobike acquisition figures

Disclaimer: These notes are shared for informational and educational purposes only. THIS IS NOT INVESTMENT ADVICE.

These notes were composed on March 18, 2019 for the China Tech Investor podcast episode 18. Meituan Dianping is traded under 03690.HK

The Market didn’t like their results, from HKD 57 down 8.3% day after they announced, fell again the second & 3rd day, for a total of -15.2%.
Since then bounced back hard, now at HKD 55.10


  • Selling & marketing expense above Gross Profit
  • Selling & marketing grew 45%, Revenues grew 92%, Gross Profit grew 23.7%
  • HUGE fair value change in convertible redeemable preferred shares, HKD 104.6 Billion, on absolute basis that’s 60% larger than revenue.
    • It is a NON-CASH item, but does create dilution, so goes through Income Statement in order to end up on equity side of balance sheet.
    • Note 14 of their filing explains Convertible redeemable preferred shares

Mobike Acquisition Figures (April 2018):

  • Consideration: RMB 9.4 B in cash, RMB 5.8 B in preferred shares
  • Valued Trade name at RMB 1.6 Billion
  • User list valued at RMB 840 M (assuming RMB 200 deposit per user about 40.6 M users = RMB 20 per user)
  • Advance from transacting users: almost RMB 500 M
  • DEPOSITS from transacting users: RMB 8.1 B (offered refunds July 2018)
    • Assuming all Meituan deposits from transacting users related to Mobike: paid back RMB 4.8B
    • Currently have RMB 3.3B on balance sheet, so checks out
  • 59% of users got their deposit back
  • Cash & cash equivalents: RMB 834 M
  • Negative Net tangible asset value: RMB -835 M
  • Revenue contributed by Mobike (from April) was RMB 1.5 B
  • Mobike loss of RMB 4.55 B
  • Using their pro forma estimates of revenue and net loss if Mobike had been consolidated from Jan 1 2018:
  • RMB 1.76 B in revenues
  • RMB -6.1 B in net loss

Cash Position:

  • Cash & equivalents = RMB 17 B
  • ST investments = RMB 41.8 B
  • Total RMB 58.8 B (unrestricted cash)
  • Cash used in operations:
    • Big item: convertible redeemable preferreds RMB 104 Billion
    • D&A: RMB 5.4B
  • Working capital
    • Decrease in deposits: RMB 4.8B
    • Increase in current assets RMB 3.7B
    • Increase in trade payables RMB 2.1B
    • Decrease in payables to merchants RMB 1.8B
  • Cash used in investing activities:
    • RMB 7.3B business combinations
    • RMB 2.2B PP&E
    • RMB 1.6B purchase of investments

Free Cash Flow Positive or burning cash?

  • I’d use Operating Cash Flow (-9.18B) minus Capex (PP&E) gives burning roughly RMB 11.4B
  • If we include business combinations, they burned RMB 18.7B
  • With RMB 58.8 B on the balance sheet, that gives about 5 years at the current cash burn rate, excluding business combinations

Meituan Expenses by nature

Thanks for reading. If you want to be notified of new posts, please subscribe. I welcome your thoughts and comments. Continue the conversation in the comments, on Twitter @jameshullx or on LinkedIn.