The Rise of China's Supercities: New Era of Urbanization

M BLUEPAPER October 10, 2019 08:00 PM GMT

China The Rise of China's Supercities: New Era of Urbanization

e believe Urbanization 2.0 will fuel productivity growth, allowing China to attain high-income status. By 2030 we expect the average size of the country's five Wsupercities to reach 120mn, an 8.5x increase in commuter rail length, and a tripling of the IoT and data market to almost US$1trn.

Morgan Stanley does and seeks to do business with companies covered in Morgan Stanley Research. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of Morgan Stanley Research. Investors should consider Morgan Stanley Research as only a single factor in making their investment decision. For analyst certification and other important disclosures, refer to the Disclosure Section, located at the end of this report. += Analysts employed by non-U.S. affiliates are not registered with FINRA, may not be associated persons of the member and may not be subject to NASD/NYSE restrictions on communications with a subject company, public appearances and trading securities held by a research analyst account. M BLUEPAPER Contributors MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Robin Xing Gary Yu Jenny Zheng, CFA Economist Equity Analyst Economist +852 2848-6511 +852 2848-6918 +852 3963-4015 [email protected] [email protected] [email protected]

MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY TAIWAN LIMITED+ Shawn Kim Laura Wang Sharon Shih Equity Analyst Equity Strategist Equity Analyst +852 3963-1005 +852 2848-6853 +886 2 2730-2865 [email protected] [email protected] [email protected]

MORGAN STANLEY TAIWAN LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Charlie Chan Tim Hsiao Jack Yeung Equity Analyst Equity Analyst Equity Analyst +886 2 2730-1725 +852 2848-1982 +852 2239-7843 [email protected] [email protected] [email protected]

MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Kevin Luo, CFA Jenny Jiang, CFA Grace Chen Equity Analyst Equity Analyst Equity Analyst +852 2239-1527 +852 2848-7152 +852 2848-5835 [email protected] [email protected] [email protected]

MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Elly Chen Lillian Lou Simon H.Y. Lee, CFA Equity Analyst Equity Analyst Equity Analyst +852 3963-0122 +852 2848-6502 +852 2848-1985 [email protected] [email protected] [email protected]

MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Richard Xu, CFA Rachel L Zhang Sheng Zhong Equity Analyst Equity Analyst Equity Analyst +852 2848-6729 +852 2239-1520 +852 2239-7821 [email protected] [email protected] [email protected]

MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Sean Wu Qianlei Fan, CFA Jack Lu Equity Analyst Equity Analyst Equity Analyst +852 3963-0755 +852 2239-1875 +852 2848-5044 [email protected] [email protected] [email protected]

MORGAN STANLEY ASIA LIMITED+ MORGAN STANLEY ASIA LIMITED+ Praveen K Choudhary Zhipeng Cai Equity Analyst Economist +852 2848-5068 +852 2239-7820 [email protected] [email protected] M BLUEPAPER Contents

5 Preface 91 2d. Banks

6 Key Charts at a Glance 97 2e. Insurance

9 Executive Summary 102 2f. Agribusiness

18 China's Path to Urbanization 2.0 106 Investment Theme #3: New Lifestyles in Smart Supercities 27 Initiative #1: City Clusters 107 3a. Transportation 34 Initiative #2: Smart Cities 111 3b. China Property 52 Initiative #3: Agricultural Modernization 117 3c. Hong Kong Property Companies 58 Investment Theme #1: From a Consumer to an Industrial Internet 121 3d. Materials

59 1a. Telecoms 125 3e. Consumer IoT

63 1b. Internet 132 3f. Education

71 1c. Tech Hardware and Software 139 3g. Healthcare

78 Investment Theme #2: Digitalization of Old- 141 3h. Macau Gaming Economy Industries 144 3i. Tourism 79 2a. Autos 148 Summary of Stocks Exposed to Urbanization 2.0 83 2b. Logistics

88 2c. Utilities and Power Equipment

MORGAN STANLEY RESEARCH 3 M BLUEPAPER Other Contributors

Equity Strategy Fran Chen Morgan Stanley Asia Limited+ Strategist [email protected] +852 2848-7135 Telecoms and Communication Yang Liu Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-1911 Camille Xu Morgan Stanley Asia Limited+ Research Associate [email protected] +852 3963-0692 Sara Wang Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-1230 Internet Eddy Wang, CFA Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-7339 Alex Poon Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 3963-3838 Steven Tsai Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2848-7217 Alex Ko Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-1225 Technology Yunchen Tsai Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848-5636 Howard Kao Morgan Stanley Taiwan Limited+ Equity Analyst [email protected] +886 2 2730-2989 Ray Wu, CFA Morgan Stanley Taiwan Limited+ Equity Analyst [email protected] +886 2 2730-2871 Daniel Yen, CFA Morgan Stanley Taiwan Limited+ Equity Analyst [email protected] +886 2 2730-2863 Lily Chou Morgan Stanley Taiwan Limited+ Research Associate [email protected] +886 2 2730-2869 Daisy Dai, CFA Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2848-7310 Jeff Hsu Morgan Stanley Taiwan Limited+ Research Associate [email protected] +886 2 2730-2864 Tony Shen Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2848-5657 Autos Shelley Wang, CFA Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 3963-0047 Frank Wan Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-1229 Logistics JunYi Yu Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-7817 Utilities and Power Equipment Eva Hou Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848 6964 Banks Anil Agarwal Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848-5842 Katherine Liu Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-1924 Lu Lu Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-1568 John Cai Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-1885 Joey Xu Morgan Stanley Asia Limited+ Research Associate [email protected] +852 3963-0337 Insurance Green Cai Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2848-5686 Birlina Qi Morgan Stanley Asia Limited+ Research Associate [email protected] +852 3963-4087 Capital Goods and Construction Hangjie Chen Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848-7168 China Property Chloe Liu Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848-5497 Cara Zhu Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2848-7117 Ziya Zhou Morgan Stanley Asia Limited+ Research Associate [email protected] +852 3963-0322 Consumers Hanli Fan, CFA Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 3963-1017 Education Elsie Sheng Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 3963-0475 Healthcare Yolanda Hu Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848-5649 Laurence Tam Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-1753 Ethan Ding Morgan Stanley Asia Limited+ Research Associate [email protected] +852 3963-0546 Alexis Yan Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-7953 Materials Sean Xiang Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2848-8154 Hannah Yang Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-7079 HK Real Estate and Macau Gaming Hildy Ling Morgan Stanley Asia Limited+ Equity Analyst [email protected] +852 2239-7834 Dan Xu Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2239-1227 Gareth Leung Morgan Stanley Asia Limited+ Research Associate [email protected] +852 2848-7339

4 M BLUEPAPER Preface

Can China maintain a relatively fast pace of growth amid the chal- safer than ever, while electric vehicles and green energy sources lenges of slowing globalization and aging demographics? Since we reduce pollution. This will enhance the population capacity of cities. last wrote in-depth on China's journey to a high-income economy in November 2017, globalization has been hampered by tariffs and What could this mean for the economy? other barriers. As the working-age population declines, demographic headwinds will also continue to exert a drag on economic growth. We We estimate that Urbanization 2.0 will attract an additional 220mn believe the answer to these challenges is a new phase of urbanization city dwellers by 2030 (vs. existing urban residents of 831mn). Half of with the potential to create productivity gains by facilitating the freer them will settle in the top five superclusters, which we project will movement of enterprises and workers while generating synergies have populations of 120mn on average – close to the size of Japan's between diverse industries. current population – enabled by an 8.5x increase in commuter rail length. The number of megacities with populations larger than New The path to Urbanization 2.0 York City (8mn) should rise from nine now to 23 by 2030. Next-gen technologies, enabled by our estimated US$800bn capex in digital In this report we look at what the next decade could hold for China's infrastructure, should sustain total factor productivity growth at cities and make our best attempt to identify investment opportuni- 1.6% annually through 2030. Offsetting industrial automation, voca- ties. We forecast that the country's urbanization rate will rise to 75% tional training should help match skilled workers with high value- by 2030, driven by (1) the growth of city clusters , which will bring added manufacturing and service jobs. Labor productivity will the benefits of urban agglomeration while alleviating big-city prob- almost double, we estimate, with 55% of the increase coming from lems, (2) smart cities that leverage next-generation technologies to the agglomeration effects of supercities and 40% from rural-urban reduce traffic, crime and pollution, and (3) agricultural migration. Most importantly, we remain confident that China will modernization , which will boost labor productivity and enable reach high-income status as annual per-capita income almost dou- more rural workers to migrate to cities. bles from US$9,450 today to US$17,800 by 2030.

Aiming for faster, safer, greener, more livable cities What could this mean for investment?

China has experienced unprecedented urbanization and economic We identify three key investment opportunities from Urbanization growth in just a few decades. In its next stage of development, how- 2.0 as disruptive technologies unleash further growth potential: ever, China is focusing on making cities faster, safer, greener, and more livable by embracing structural reforms (such as Hukou and l Transition from consumer to industrial internet: 5G infra- land reforms) and a new era of digital technologies. By 2030, city structure, public cloud, Internet of Things (IoT) devices, and residents should generally be able to reach their workplaces within software. 15 minutes. At home, current plans aim to have 5G-enabled smart l Digitalization of old-economy industries: Electric and auton- appliances clean, cook, and order food when supplies run low, while omous vehicles, smart grid and utilities, market-oriented banks virtual reality headsets will help students do homework and attend and insurers, and smart farming. online tutoring classes with the country's top teachers. l New lifestyles in smart supercities: E-commerce, smart home appliances, online tutoring and vocational education, health- Much of this is a reality today. It is already common to pay at grocery care, and railway construction. stores in Hangzhou or check in at the new Beijing airport using only facial recognition. By 2030, greater changes will be enabled by Although the views in this report are our base-case scenarios, we investments in digital infrastructure and the adoption of artificial acknowledge that there is no way to foresee all risks that may arise intelligence (AI) and big data. We expect high-speed commuter trains, from issues like automation, big data, and trade tensions, which could smart traffic control systems, shared mobility, and automated include job losses, a higher debt burden, or other unintended social vehicle technologies to cut travel times and make streets and roads issues. For more on risks, please see Where could we be wrong?

MORGAN STANLEY RESEARCH 5 M BLUEPAPER Key Charts at a Glance

Exhibit 1: Forecasts for Urbanization 2.0

Source: Morgan Stanley Research estimates

6 M BLUEPAPER Exhibit 2: From Urbanization 1.0 to 2.0

Source: Morgan Stanley Research

Exhibit 3: Did you know…

Source: NBS, Haver, Morgan Stanley Research

MORGAN STANLEY RESEARCH 7 M BLUEPAPER Sector Implications Through 2030 Top Stocks for This Theme (1) From a consumer to an industrial internet • 5G capex to reach US$400bn in 2019-30, 2x that of 4G • China Tower (0788.HK) Telecoms • Key beneficiaries: 5G infrastructure companies • Alibaba (BABA.N) • GigaDevice Semiconductor Beijing • Enterprise IT spending on software and IT services to account for around 8% of global spending (vs. 3% now) Internet • Key beneficiaries: Technology leaders that are expanding from consumer to industrial applications (603986.SS) • HIKVision Digital Technology • IoT device market size to more than double to US$700bn (002415.SZ) Tech Hardware and • Software and IT services market size to fivefold to US$200bn • Yonyou Network Technology Software • Key beneficiaries: Top players geared to IoT and 5G; software vendors focusing on digital transformation or with (600588.SS) smart city exposure • VenusTech (002439.SZ) (2) Digitalization of old-economy industries • Shared mobility to take up 10% of total car parc (vs. 2% now) • Market share of electric vehicles to reach one-third (vs. 4% in 2018) • S.F. Holding (002352.SZ) Autos • Share of cars with high/full automation to reach 20% (vs. 0% now) • Key beneficiaries: Early movers in EVs and autonomous vehicles • Logistics cost/GDP to come down to 10% (vs. 15% in 2018) • NARI Technology (600406.SS) • Nationwide delivery time to be within 1 day (vs. 2-4 now) Logistics • Express volumes to reach 300bn deliveries per year (up 6x) • Key beneficiaries: Logistics companies with strong R&D investment

• Share of clean energy in capex to rise to 60% (vs. 40% today) • Ping An Bank (000001.SZ) Utilities and Power • Share of clean energy in power generation approaching 40% (vs. 30% now) Equipment • Capex in smart grid to increase 2.64x, to US$80bn, in 2021-30 • Key beneficiaries: Utility players with competitive edge in smart grid • Healthier credit growth, at 7% CAGR (vs. 17% in the past decade) • Ping An Insurance Company (2318.HK) Banks • Key beneficiaries: More market-oriented banks • Insurance penetration to rise to 9% (vs. 4.3% in 2018) Insurance • Key beneficiaries: Insurers with leading positions in top-tier cities and advanced technological capabilities • Yuan Longping High-tech Agricultural • GM corn and soybean seed application to reach 50% (vs. 0% today) (000998.SZ) Agribusiness • Key beneficiaries: Agribusiness entities with strong brand name and GM seed pipeline (3) New lifestyles in smart supercities • High-speed rail length to reach 65,000km (vs. 30,000km now) • CRRC Corp Ltd (1766.HK) Transportation • Inter-city commuter rail to reach 17,000 km (vs. 2,000km now) • Key beneficiaries: Railway construction companies focusing on inter-city and metro rail build-up • Annual incremental housing demand to sustain, at 1,450mn sqm (vs. 1,479mn sqm in 2018) Property • Annual housing price growth to reach 6% in five key city clusters (vs. 4% elsewhere) • Haier Smart Home (600690.SS) • Key beneficiaries: Developers with more landbank exposure to large cities and key cityclusters • Market share of top ten players in steel and cement to reach 60% (vs. 37% now) and 70% (vs. 57% now) respec- Materials tively • Key beneficiaries: Leaders in highly concentrated industries • Meituan Dianping (3690.HK) • 100% penetration of smart home appliances (vs. 20% today) • IoT devices to reach 7 units per households (vs. 1 today) Consumption • E-commerce penetration to exceed 75% of total population (vs. 44% today) • New Oriental Education & Technology • Key beneficiaries: Companies with clear strategies for smart home appliances and e-commerce leaders Group (EDU.N) • Penetration of K-12 online tutoring to exceed 35% (vs. under 10% now) Education • Vocational training market to triple, to US$300bn • Key beneficiaries: Top online tutoring and vocational education players • China's pharmaceutical market to reach US$0.5trn by 2030 (6.3% CAGR in 2018-30) • TAL Education Group (TAL.N) • China's healthcare service market to reach US$2.2trn by 2030 (10.1% CAGR in 2015-30) Healthcare • Key beneficiaries: Pharmas with strong, innovative pipelines, and top healthcare service providers with solid growth potential • Jiangsu Hengrui Medicine (600276.SS) • Gaming revenue to more than double, to US$70-100bn Macau Gaming • Key beneficiaries: Gaming operators in Macau with bigger hotel capacity • Domestic annual tourism expenditure to reach US$1.5trn (vs. US$0.78trn in 2018). Tourism • Key beneficiaries: Top tourist destination operators • Aier Eye Hospital (300015.SZ)

8 M BLUEPAPER Executive Summary

China is answering an ancient question...... with city clusters and smart cities Source: Shutterstock Source: Shutterstock

Philosophers, policymakers and academics have debated the Launching China's second phase of optimum size of cities since Aristotle walked the streets of ancient Athens. Simply put, cities that are too small lack the labor and effi- urbanization ciencies to power economic growth, and those that are too large become expensive to manage as efficiencies break down under the Why continued urbanization is important: China’s productivity weight of overcrowding and decaying infrastructure. growth has almost halved, to 2.3% annually in 2010-18 from 4.4% in 2000-09. We believe the way to address this is through continued China obviously has no problem with cities being too small. But urbanization, which can provide economic benefits such as improving where does it stand with respect to the latter problem now that 60% the ease with which enterprises and workers can move to productive of the population lives in urban areas, up from a mere 18% in 1978? locations to facilitate matching, sharing and learning, spread ideas, And what is the outlook for urbanization as China’s population ages form specialized supply chains, and generate synergies across dif- and starts to decline in absolute terms? ferent sectors. Overcoming hurdles to urbanization will require structural reforms, including the freer flow of labor, better social A well-worn concern is that 'China will grow old before it gets rich'. welfare coverage, and larger-scale farming. Meanwhile, technology While it is true that demographics will be a significant drag on eco- can also help enhance liveability in densely populated cities and nomic growth in the years ahead, we argue that it will not stop the boost productivity growth. process of urbanization. On the contrary, we see urbanization as a solution to China's demographic pressures that will lift the economy From Urbanization 1.0 to 2.0: Despite the substantial economic to high-income status. In the shorter term, urbanization-related success of urbanization as China initiated reforms and opened up investment and productivity gains should figure as a key feature of over the past four decades, certain bottlenecks are showing the Beijing's counter-cyclical stimulus, helping ease the effects of limits of the old growth model. These include (1) big-city problems US-China trade tensions. such as traffic congestion, social issues and pollution, (2) policy constraints on labor mobility, (3) a shrinking pool of rural manpower for further urbanization, and (4) increased trade tensions with the US.

MORGAN STANLEY RESEARCH 9 M BLUEPAPER To reach the next stage of sustainable development, China is forging The transition toward Urbanization 2.0 is also evident in a shift a different urbanization path, focusing on making cities faster, safer, in policy focus: In our view, the urbanization strategy over the past greener, and more livable by embracing structural reforms and a new two years has been shifting to focus more on promoting mega-clus- era of digital technologies. We see three initiatives underpinning ters in advanced regions through a combination of (1) top-down initia- Urbanization 2.0: tives (such as Special Districts) to strengthen coordination across local administrative boundaries within clusters, and (2) more market- 1. City clusters , knitted together by the country's advanced oriented approaches to avoid inefficiencies from limited demand for rail system, should continue to reap the benefits of urban infrastructure and services in less populated, remote inland regions. agglomeration while alleviating big-city problems. We This would be distinct from past regional rebalancing initiatives believe five key city clusters across the country – Yangtze (including 'Western Development' since 2000 and 'Northeast River Delta, Jing-Jin-Ji Area, Greater Bay Area, Mid-Yangtze Revitalization' since 2004), which were mainly aimed at reducing River Area, and Chengdu-Chongqing Area – will likely regional income gaps and relieving pressure from population inflows account for 75% of GDP growth and half of the urban pop- into developed coastal regions. ulation increase in 2019-30. 2. Smart cities that leverage technologies like 5G, cloud, big As evidence of this, the August 2019 meeting of China's Central data, the Internet of Things (IoT), and artificial intelligence Economic and Financial Affairs Commission – the country’s highest (AI) should help reduce traffic, crime, and pollution, and economic policymaking body – specifically mentioned that "hub improve the quality of city life, greatly enhancing the cities and city clusters are becoming the main medium for growth and capacity of the cities of tomorrow. We expect the number development" and that policies should enhance the economic and of megacities with populations similar to or larger than population capacity of these regions to facilitate the agglomeration New York City (8mn) will reach 23 by 2030 (vs. 9 today). of productive factors. This marked the first time in many years that 3. Agricultural modernization through land reforms and policymakers have emphasized the role of large city clusters in the wider adoption of smart farming should boost labor advanced regions. Meanwhile, we have seen strong policy support productivity, enabling more rural workers to migrate to for the three initiatives in the past two years. cities. We expect China's agricultural labor productivity to more than double over the next decade, freeing up more of the rural population for further urbanization.

Exhibit 4: Three key initiatives for China's Urbanization 2.0

Source: Morgan Stanley Research

10 M BLUEPAPER

Source: AlphaWise, Morgan Stanley Research

Exhibit 5: Strong government support for Urbanization 2.0

Source: Xinhua, Morgan Stanley Research

MORGAN STANLEY RESEARCH 11 M BLUEPAPER

In our view, China is poised to be the global leader Exhibit 6: in smart city and city cluster development. This is China's unique advantages in Urbanization 2.0 underpinned by: Extensive High-Speed Rail Network l The world's longest and fastest high-speed km/h – The World's Fastest rail (HSR) system, helping realize a 'one-hour 350 living cycle' in city clusters. 1 Hour Living Circles in City Clusters l Advanced 5G planning for an industrial Ahead-of-the-Curve 5G Planning Internet of Things and high e-commerce penetration, which have laid a solid foundation for 5G capex of US$400bn, 2x 4G smart city development. l Strong human capital, with millions of gradu- 300+ prefecture cities covered by 2020 ates and rapid developments in vocational High Penetration of E-commerce and Mobile Payment training meeting demand for talent in higher value-added manufacturing and service sectors. 18.4% E-retail Penetration Rate US$23bn E-hailing Market Size A bright outlook for 86% Mobile Payment Penetration Rate urbanization through 2030 Stronger Human Capital

Near-term counter-cyclical easing to boost 109mn Average Population in 5 City investment needed for Urbanization 2.0: The pre- Clusters, which is...... vailing US-China trade tensions are bringing down- ~5x that of New York ~3x that of Greater Tokyo side risks to growth trajectories both globally and in China (see Global Macro Briefing: Inching Closer to a Annual Average No. No. of International R&D Expenditure of College Graduates Patents Global Recession, 25 August 2019). In response, we expect China's policymakers to step up counter-cy- China*: 11.6mn US: 56,142 US: US$513bn China*: US$297bn clical easing, with a focus on boosting urban invest- US: 3.7mn China: 53,345 Japan: 49,702 Japan: US$155bn ment. Policy guidance at the 3Q Politburo and State Japan*: 1.0mn UK: 0.8mn Germany: 19,883 Germany: US$103bn Council meetings on September 4 suggested that Korea: 0.6mn UK: 5,641 Korea: US$60bn infrastructure investment will be aimed at building Source: Analysys Mason, Euromonitor, PWC Global Consumer Insight Survey 2019, CEIC, Haver, Morgan Stanley Research. Note: up city clusters (commuter rail), renovating urban Data as of 2018 unless otherwise noted. Annual average number of college graduates: 2013-2017 data for China and 2012-2016 data for the other countries. 2013-2016 data for Japan due to missing data in 2012, R&D Expenditure: 2018 data for China and 2016 data facilities (car parks, cold chains for food logistics), for the others. 2018 data for the others. and next-gen mobile networks (5G), which are essential investments for Urbanization 2.0.

We expect China's urbanization ratio to reach 75% by 2030, up from 60% today, translating into 220mn new urban dwellers. We expect total factor productivity to be sustained at a 1.6% CAGR through 2030 (vs. 1.9% in 2014-18), as labor productivity increases by 80% from today's level. Our growth accounting analysis suggests that 55% of the labor productivity increase will come from the agglomeration effect of smart supercities, with another 40% attributable to rural-urban migration and 5% coming from agricultural modernization on the back of land reforms and smart farming. We thus remain confident that China is poised to attain high-income status by 2025, with annual per capita income almost doubling, from US$9,450 today to US$17,800 by 2030.

12 M BLUEPAPER Manageable financial stability risk from Urbanization 2.0 buil- and continued opening-up efforts will help attract more foreign dout: Contrary to market concerns that increased capex demand for investment. The potential for foreign inflows into China’s equity and Urbanization 2.0 will lead to renewed debt problems, we believe this bond markets will provide support to China’s overall balance of pay- risk is manageable considering (1) that there is less need for massive ments and the RMB exchange rate over the longer run (see EM investment given China’s strong foundation of infrastructure today Strategy and Economics: The Transformation of China's Capital (for instance, we estimate the combined capex needed for digital Flows, 11 February 2019). infrastructure, high-speed rail and the smart grid – three key components of the smart supercity buildout – will be less than US$200bn per year in 2019-30, only about 10% of China's annual infrastructure Near-term catalysts and milestones of FAI in the past five years); (2) more transparent funding given local Urbanization 2.0 governments' shift from shadow bank borrowing to bond issuance and a potential increase in private investment with market-oriented reforms; and (3) relatively higher asset quality in digital infrastruc- We expect continued Hukou and land reforms, the rapid develop- ture and HSR (which will concentrate more on eastern China and ment of high-speed inter-city rail, electric vehicles, shared mobility, inter-city commuter rail). These factors, combined with continued investments in 5G, AI and big data technologies, and rapid growth in shadow bank tightening, affirm our view that China will stabilize its vocational training to be key drivers of further urbanization over the debt to GDP ratio in the next decade (see China: Blue Paper Revisit: next 3-5 years. Over the longer run, more advanced smart city fea- Why we are still bullish on China, 14 November 2017). tures, such as driverless cars, auto-delivery drones, and fully interconnected and automated home appliances should take productivity More opportunities for foreign capital: We believe China’s growth to the next level. To track the progress of China's improving project funding structure, higher-quality investment Urbanization 2.0, we will look at several markers over the next 3-5 opportunities in industry digitalization and smart city development, years:

Exhibit 7: Near-term catalysts to watch

Source: NDRC, State Council, Xinhua, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 13 M BLUEPAPER What does Urbanization 2.0 suggest for The transition to Urbanization 2.0 should open up significant investment opportunities across sectors. The distinction between the old industries and companies? and new economies at the sector level may not be so clear-cut any- more. Leveraging Morgan Stanley Research's sector teams, we iden- Access to Chinese equity markets has been largely synchronous with tify three key investment themes across 16 industries to create a list the country's urbanization and globalization processes, offering of the top 18 stocks that can provide exposure to Urbanization 2.0. unique investment opportunities at different stages to foreign investors. 1. From a consumer to an industrial internet: We expect a major transition from consumer-based to industrial-based l Stage 1: China joins global economy and top-down, policy- applications, given that consumer adoption is already very driven infrastructure investment benefits Old Economy* mature. With strong government support for the 5G sectors: Key defining events of this period include China re-en- rollout (one of the major enablers), key 5G infrastructure tering the WTO in 2001, the opening of A-share markets to for- companies will benefit. Meanwhile, the development of eign investors through the QFII program in 2002, and the IPOs the industrial internet will help support leading public of major SOEs, such as PetroChina, Sinopec and CNOOC in cloud companies. As we expect the market size of IoT 2000-01, and the Big Four banks starting from 2005. devices and software to triple by 2030, we believe that l Stage 2: Transformation to a more consumption-driven Asian tech players geared to IoT and 5G are poised to gain. model, with New Economy** focused investments paying 2. Digitalization of old-economy industries: We expect off: This period is defined by heightened debt levels and ROE early movers in electric and autonomous vehicles, logistics deterioration in SOE-dominated cyclical spaces, as well as the companies with strong R&D investment, and utility players fast growth of the internet-led New Economy (Tencent with a competitive edge in the smart grid to gain the most becoming the largest Hong Kong listed stock by market capital- from the development of smart cities. In the financial ization, and Chinese ADRs getting included in MSCI indices in sector, front-runners will likely include market-oriented 2016). banks and insurers with leading positions in top-tier cities and advanced technological capabilities. Smart farming *Old Economy: Materials, Energy, Industrials should also benefit agribusiness entities with leading position and efficient products. **New Economy: Consumer (Staples + Discretionary), Media & 3. New lifestyles in smart supercities: Digitalization is on Entertainment, IT, Healthcare course to change household lifestyles, benefiting e-commerce leaders and companies with a clear strategy for Exhibit 8: smart home appliances. Meanwhile, continued population The investment-driven Old Economy outperformed in Stage 1 and the inflows into supercities should support top online tutoring consumption-driven New Economy is outperforming in Stage 2 and vocational education players, as well as healthcare 180 companies with good hospital networks and advanced New Economy Rel. to Old Economy 160 technologies. We also expect railway construction compa- 140 nies focusing on inter-city rail to benefit from the govern- 120 ment's support for city clusters. 100

0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Source: FactSet, Morgan Stanley Research. Data as of end-September 2019. MSCI China sector level sub-indices are used to construct performance index for New Economy (Consumer Staples, Consumer Discretionary, Media & Entertainment, IT, Healthcare) and Old Economy (Materials, Energy, Industrials.

14 M BLUEPAPER

Exhibit 9: Top stocks exposed to the Urbanization 2.0 theme

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 15 M BLUEPAPER

Exhibit 10: Chinese unicorns that could benefit from Urbanization 2.0 Company name Valuation (US$bn) Industry Investors Toutiao (Bytedance) 75 Artificial intelligence Sequoia Capital China, SIG Asia Investments, Sina Weibo, Softbank Group Didi Chuxing 56 Auto & transportation Matrix Partners, Tiger Global Management, Softbank Corp., Kuaishou 18 Mobile & telecommunications Morningside Venture Capital, Sequoia Capital, Baidu Bitmain Technologies 12 Hardware Coatue Management, Sequoia Capital China, IDG Capital DJI Innovations 10 Hardware Accel Partners, Sequoia Capital Guazi (Chehaoduo) 9 E-commerce & direct-to-consumer Sequoia Capital China, GX Capital EasyHome 5.7 Consumer & retail Alibaba Group, Boyu Capital, Borui Capital GuaHao (We Doctor) 5.5 Health Tencent, Morningside Group Hello TransTech 5 Auto & transport Ant Financial Services Group, GGV Capital UBTECH Robotics 5 Hardware CDH Investments, Goldstone Investments, Qiming Venture Partners United Imaging Healthcare 5 Health China Life Insurance, China Development Bank Capital, CITIC Securities International Meizu Technology 4.58 Hardware Telling Telecommunication Holding Co., Alibaba Group Vipkid 4.5 Edtech Sequoia Capital China, Tencent Holdings, Sinovation Ventures Face++ (Megvii) 4 Artificial intelligence Ant Financial Services Group, Russia-China Investment Fund, Foxconn Technology Company XPeng Motors 3.65 Auto & transportation Morningside Venture Capital, Foxconn Technology Company, Alibaba Group Cloudwalk 3.32 Artificial intelligence Oriza Holdings, Guangdong Technology Financial Group Huitongda 3.18 E-commerce & direct-to-consumer Alibaba Group, Shunwei Capital Partners, New Horizon Capital Horizon Robotics 3 Artificial intelligence Hillhouse Capital Management, Linear Venture, Morningside Venture Capital Xiaohongshu 3 E-commerce & direct-to-consumer GGV Capital, ZhenFund, Tencent

Source: Morgan Stanley Research, CB Insights

Unlisted companies starting to thrive in the new era of urban l Government-led investment in city clusters and smart cities development: Urbanization 2.0 offers significant opportunities in could result in debt problems. Investment without coordi- terms of market growth, penetration and consolidation as well as nated regional planning and detailed risk-reward analysis could new technological applications, in our view. We note that a number lead to insufficient utilization of infrastructure connectivity, of unlisted Chinese companies have developed leading positions in inefficient debt buildup and wasted resources, hampering related industries. China's productivity growth (see a more detailed discussion here ). l Where could we be wrong? Land and Hukou reforms may stall. We believe land reforms hold the key to boosting large-scale farming and freeing up more of the rural population for urbanization, while Hukou Although our base-case view is for the continued development of reforms will help migrant workers gain access to the social China's smart cities and city clusters, we acknowledge that there is no security system and have a greater sense of belonging in cities. way to identify all risks or unintended consequences that could occur Hence, slower-than-expected reforms on these fronts could in the years to come. Having said that, some of the key risks that could exert a drag on further urbanization. deter the progress of Urbanization 2.0 include: l Privacy concerns regarding big data: As mentioned, one advantage China has in developing smart cities is fewer hurdles l Unintended social issues: The wide adoption of automation in consumer data collection. However, should social concerns and AI could lead to larger-than-expected job losses, particu- over data privacy increase significantly, it may slow the pace of larly in construction and lower value-added manufacturing sec- new technology adoption, which relies heavily on big data anal- tors. While we believe vocational training and China's ysis. developing service sector could help mitigate this problem, fric- l Tech supply chain decoupling: Currently, the global tech tional unemployment could still increase if high-tech adoption value chain is still highly dependent on US components, partic- takes place at a faster pace. Meanwhile, a rise in capital produc- ularly high-end semiconductors such as 5G- and AI-related tivity as new technologies are introduced may lead to more chips. The US government included Huawei and Hikvision on its severe income disparities, adding to the urgency of social secu- Entity List in May and October, respectively, raising concerns rity and welfare reforms (we discuss factors that could mitigate that chip supply could lead to delays in 5G and/or smart city these effects here ). development in China (see a more detailed discussion here ).

16 M BLUEPAPER

Macro Analysis

MORGAN STANLEY RESEARCH 17 M BLUEPAPER China's Path to Urbanization 2.0

Overview Defying structural headwinds from slowing globalization and aging demographics, we believe China's next leg of urbanization – underpinned by highly productive city clusters, smart cities, and agricultural modernization – will help sustain productivity growth, lifting the country to high-income status by 2025.

Key forecasts – Urbanization ratio to reach 75% (60% in 2018), bringing in 220mn new city-dwellers

– Total factor productivity to be sustained at a 1.6% CAGR (1.9% in 2014-18), with labor productivity almost doubling

– Nominal GDP per capita to almost double to US$17,800 (from US$9,450 today) as China exceeds the high-income threshold by 2025

l From Urbanization 1.0 to 2.0 Shrinking rural manpower. The rural population is aging rapidly. Combined with slower growth in migrant workers since 2010, this has raised questions about how much more urban- Since beginning its 'reform and opening up' process, China has ization is possible. learned from developed nations and has achieved outsized l Trade protectionism has increased. The escalation in gains: Supported by favorable demographics, a high savings rate and US-China tariffs over the past two years has pressured manu- the globalization of supply chains, China has experienced unprece- facturing employment, leading to worries that supply chains dented progress in urbanization. The urban population has grown could migrate away from China over the long term if tensions fivefold, to nearly one billion people, from 172mn in 1978, fueling persist. This could shrink manufacturing job opportunities for China's rise to middle-high income status as it became a hub of global migrant workers, particularly in lower value-added sectors. manufacturing. With 33 cities of 5mn or more, China has far outpaced the US, which has just 9 metropolitan areas at this level today. We In our view, China's next phase of urbanization seeks to over- call this phase Urbanization 1.0. come these obstacles through three initiatives – city clusters, smart cities, and agricultural modernization. While the closely However, the market is increasingly concerned that China's linked city clusters should continue to reap the benefits of urban urbanization is reaching its limits, in view of the following bottle- agglomeration while alleviating big-city problems, advanced tech- necks: nology and data-driven smart cities can be optimized to accommodate larger populations. Meanwhile, agricultural modernization on l 'Big-city diseases' are on the rise, including traffic congestion, the back of land reforms and the wider adoption of smart farming crime, pollution, high property prices, and inadequate educa- should boost labor productivity, enabling more rural workers to tional and healthcare resources. migrate to cities (see detailed discussion in the next three chapters). l Policy constraints on labor mobility. Hukou (city resident In our view, these initiatives will be underpinned by digital infrastruc- permit) restrictions in larger cities and a fragmented social ture and continued Hukou and land reforms. security system have resulted in insufficient insurance coverage for migrant workers.

18 M BLUEPAPER China's unique advantages so far focused on consumer applications like fixed wireless and smartphones. Meanwhile, China's mobile payment penetration is the highest in the world at 86%, vs. a global average of 34%, as it leads In our view, China is poised to be the global leader in smart city the way toward a cashless society. and city cluster development. China already has the world's longest and fastest high-speed rail system. As an example, it only takes As China is an early adopter of new technologies and faces fewer hur- 45-60 minutes to travel from Shanghai to Hangzhou by HSR (number dles to doing so than its Western counterparts when it comes to of trains per day: 80) – much faster than the 80-120 minutes to go issues such as consumer data collection, its innovators have a signifi- from London to Birmingham (number of trains per day: 65), which is cant advantage in big data and AI, which could eventually expand a comparable distance. We expect the length of the high-speed rail today's digital payment and e-commerce ecosystems to connect network to increase to 65,000km by 2030, up from 30,000km every aspect of consumers' daily lives via the Internet of Things. today, within which inter-city commuter rail length would increase 8.5x, to 17,000km. This would realize a 'one-hour living cycle' in city On the human capital front, China's large domestic talent pool has clusters and reduce logistics costs and delivery times. averaged 11.6mn graduates annually over 2013-18, far exceeding levels in developed countries, and young people are graduating from Meanwhile, China's advanced 5G planning for an industrial Internet of college at a rapidly increasing rate ( Exhibit 14 ). Meanwhile, rapid Things and high e-commerce penetration have laid a solid foundation developments in vocational training and online education should for smart city development. Albeit not the first to launch globally, help match migrant workers with jobs in higher value-added manu- Chinese telcos have emphasized industrial IoT applications, including facturing and service sectors. smart cities, as opposed to the US and Korea where operators have

Exhibit 11: Exhibit 12: Significantly shorter travel times in China's key city clusters than other 5G – The key building block for smart cities developed countries (Hrs) Comparing Travel Time Between Major Cities 3 2.5 2.0 2 1.25 1.25 1.0 0.75 1 0.5 0.25 0 (Train) (HSR) (HSR) SF - SF Palo - Alto (Shinkansen) (HSR) Tokyo- Osaka or or Train) Guangzhou - Shanghai - Hefei (Virgin Trains) Zhongshan (HSR) HK - Shenzhen CBD NYC - Stamford(Car London- Birmingham Shanghai - Hangzhou ~45KM ~60KM ~160KM ~400km

Source: Morgan Stanley Research

Source: ITU

MORGAN STANLEY RESEARCH 19 M BLUEPAPER

Exhibit 13: Exhibit 14: China's number of new graduates per year far exceeds that of key devel- More and more young people are graduating from college oped countries 40% Tertiary Graduates / Births 22 Years Ago Annual Average No. of Graduates from Tertiary Education (Thousand 35% People) 30% China* 11,591 25%

20% United States 3,733 15%

Japan* 979 10%

5% United Kingdom 768 0% 1986 1990 1994 1998 2002 2006 2010 2014 2018

Korea 621 Source: Ministry of Education, NBS, Morgan Stanley Research

Source: World Bank, Morgan Stanley Research. Note: 2013-2017 data for China and 2012-2016 data for the other countries. 2013-2016 average data for Japan because of missing data for 2012.

Exhibit 15: Exhibit 16: Higher penetration of e-commerce...... and mobile payment

South Korea 24.0% Mainland China 86% China 18.4%

UK 17.0% Hong Kong SAR 64% US 13.7%

Germany 11.0% Singapore 46% Taiwan 10.8%

Australia 10.0% Online Shopping/Total Mobile Payment Russia 45% France 8.8% Retail Sales, 2018 Penetration Rate*, 2019

Japan 8.6% Global Average 34% Switzerland 6.7%

0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 0% 20% 40% 60% 80% 100% Source: Wind, Euromonitor, Morgan Stanley Research Source: PWC Global Consumer Insight Survey 2019. *Share of mobile payment in total purchase

20 M BLUEPAPER

Envisioning the potential of urban life As a tech analyst, Mrs. Yang needs to make a site visit to a company that makes industrial robots. She sets off for in 2030 Guangzhou at 9am to visit the factory, a trip that takes less than 15 minutes by driverless flying taxi. On the taxi To illustrate how smart supercities could improve most she uses VR glasses to watch a video on the use cases for aspects of urban life, we depict the daily lives of Dr. and the company's robots. Mrs. Yang and their daughter Lily (aged 8) in 2030. The family lives in Huizhou, a satellite city next to Shenzhen. Arriving at the factory at 9:15, she tours the facility to Dr. Yang is cardiovascular surgeon and Mrs. Yang is a understand the company's competitive advantages in stock analyst at a leading investment bank. They both technology and production. At noon she takes a work in the Futian District (Shenzhen's CBD) but moved driverless taxi to a business lunch at headquarters with to Huizhou five years ago to enjoy extra living space and a management, engineers and industrial consultants. After more relaxed lifestyle. lunch the company takes her to visit key suppliers in Foshan. Mrs. Yang sees three suppliers and then uses VR The Yangs get up at around 7am. The smart kitchen has for a real-time site visit with another in Fujian Province. breakfast ready by 6:50, including congee, sandwiches, Her visit ends at 5pm, and she returns to Huizhou before fried eggs, and warm soy milk. After eating and getting 6 to pick Lily up at school. dressed, they leave home at 7:30. The couple escort Lily to school nearby and then catch the 7:55 commuter train. Dr. Yang also has a busy day. On the way to work, he They don’t need to buy tickets, as smart cameras with opens his tablet, which displays his patients’ real-time facial recognition and GPS in their phones detect where vital signs thanks to smart sensors connected to the 5G they get on and off the train and automatically bill them network. Big data analysis predicts the intra-day progress by e-payment. On the platform, a mobile app indicates of his patients’ conditions and automatically generates his which seats the station's cloud computing system has schedule based on their urgency. At the office, a double assigned them. They reach Futian at 8:20 and take short espresso is waiting on his desk, as his smart watch has walks to their offices. detected subpar sleep quality the night before and has instructed the coffee machine to ensure he gets enough caffeine before going into surgery.

VR and 5G are enabling remote surgery Source: Shutterstock

MORGAN STANLEY RESEARCH 21 M BLUEPAPER

As a prominent heart surgeon, Dr. Yang receives requests After their busy days, the Yangs want to relax. The from around the country to conduct consultations, community’s smart butler system that's connected to all attend conferences and perform surgeries. In the their home appliances via the internet can do most of the afternoon, he virtually scrubs in at an operating theater housework. On her way home, the house management in Beijing, giving verbal guidance to the surgeons. The app on Mrs. Yang’s phone automatically changes her high bandwidth and near-zero latency of 5G-connected status to 'returning home'. The smart butler receives the cameras enable Dr. Yang to observe the operation in real order and starts tracking her distance from home. To save time, allowing him to give instructions and observe any energy, it turns on the air conditioner 10 minutes before subtle changes near the incisions. she arrives and switches on the lights in the living room 2 minutes before. The app also asks for her dinner What about Lily? Her school believes learning by preferences and gives orders to the smart kitchen, which immersion is more efficient than from a textbook. In her is linked to the refrigerator and can automatically prepare science class, Lily puts on VR glasses to watch how cooking materials. The Yangs don’t need to go grocery tornadoes form, then takes a quiz. She hits the right shopping, as the refrigerator knows which foods need answers on her e-ink tablet, earning rewards points that restocking and orders them online. Flying drones deliver she can exchange for stationery. Lily likes mathematics as food directly to the refrigerator within 20-30 minutes by well. Instead of having students memorize formulas from opening a window and the refrigerator door using sensors a book, her teacher has them play interactive VR games and designated QR codes. to learn the basics. The centralized educational system The family comes home at 7pm and has dinner together. tracks each student's learning curve and designs an Afterwards, Dr. Yang clears the table and puts the optimal curriculum. When Lily does her homework on the remains into the trash bin, which has an automatic sorting tablet, for example, she has more videos to watch than function and sends the waste to the community's trash her classmate Xiaohua, as she got fewer questions right center. The family decides to take a walk. The in class. neighborhood is safe even after 9pm, as smart cameras detect anyone suspicious and remind security in the community to monitor them. By 10pm, everyone is asleep. This scenario is meant only to showcase the potential of technologies already here or in development as per what we know today.

Homework hasn't gone away in the future, but at least it's more interesting Source: Shutterstock

22 M BLUEPAPER Macro outlook toward 2030 urban agglomeration will lead to efficiency gains and knowledge spillover, propelling China’s transition to high value-added activities. Short term: Countercyclical easing to accelerate Finally, the accommodation of incremental urban residents requires investment needed for Urbanization 2.0 more investment in infrastructure and housing, which may also benefit existing urban residents. Policymakers are stepping up countercyclical easing as persistent trade tensions put pressure on growth and the labor market. In our In this context, we expect China's total factor productivity view, as muted private confidence will suppress the multiplier effect growth will sustain at a robust 1.6% CAGR up to 2030 (compared of any tax cut, additional easing will be in the form of direct public with 0-1% for most developed countries since 2005, in particular a investment. Moreover, infrastructure investment in the current 0.4% CAGR for the US and 0.6% for Japan), contributing 36% of easing cycle will be geared towards city cluster buildup and next-gen overall growth (compared with 30% since the 2008 global financial mobile networks. This is evident by recent policy moves listed in crisis). Meanwhile, we expect labor productivity (output per worker) Exhibit 17 . to rise by 80% over the forecast horizon. Our growth accounting analysis shows that 40% of the labor productivity increase is Long term: Urbanization 2.0 to sustain productivity achieved by rural residents migrating to cities, facilitated by freed-up growth rural labor and enhanced urban capacity; another 55% is attributed to productivity increases within cities, driven by rising agglomera- We expect China's urbanization ratio to reach 75% by 2030, up tion, capital investment (5G, HSR, etc.), and associated technological from 60% today, largely continuing the average pace of 1.2ppt per progress; finally, the remaining 5% is attributed to agricultural mod- year over the past five years (compared with an annual increase of ernization, as the fast rise in agricultural productivity is partly offset 0.5ppt for the US during 1880-1950 and 0.9ppt for Japan during by a (desired) reduction in rural employment. 1950-2010). This implies another 220mn rural residents will migrate to urban areas. Continued urbanization will drive China’s path to high We thus reiterate our view that China will reach high-income income via a number of channels. First, migration from rural areas status by 2025, i.e., its gross national income per capita will reach should help sustain return on capital (via increased labor supply) and US$13,900, exceeding our calculated high-income threshold for encourage the continued investment of productive capital. Second, 2025 (US$13,700) as defined by the World Bank, and pick up further aggregate productivity will be supported as labor shifts to non-farm to US$17,800 by 2030 (vs. US$9,450 today). activities that tend to have higher productivity. Third, increased

Exhibit 17: Countercyclical public investments driving the next wave of urbanization

Date Government Entities Key Announcement

The Ministry of Industry and Information June, 2019 • Granted 5G commercial licenses, enabling commercial launch as early as October Technology

• Speed up the construction of parking lots July, 2019 China's Politburo • Renovate dilapidated urban neighborhoods • Promote a new information network

• Emphasized the role of large city clusters in advanced regions for the first time in many years; Central Financial and Economic Affairs August, 2019 • Policy should enhance the economic and population capacity of these regions to facilitate the Commission agglomeration of productive factors

• Infrastructure investment should concentrate on transportation (particularly railways and car parks), September, 2019 State Council energy (power and natural gas grid), utilities, and social welfare (healthcare, education and elderly care)

The Ministry of Industry and Information • Called for the build-up of 5G network and Internet of Things to promote the quality of manufacturing and September, 2019 Technology service production

Source: Government websites

MORGAN STANLEY RESEARCH 23 M BLUEPAPER

Exhibit 18: Exhibit 19: Sustained pace of urbanization towards 2030 China's productivity growth to remain robust in a global context Urbanization Ratio, % Projected Total Factor Productivity CAGR (during 2005-17 unless otherwise specified) China 100% China 90% Japan Korea 80% 2030E: China (2019-30E) US 75% Germany 70% Japan 60% United States 2018: Australia 50% 59.6% Finland 40% Canada Denmark 30% France 20% Sweden 10% Netherlands United Kingdom 0% Belgium Italy

1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2018 -0.5% 0.5% 1.5% 2.5% 3.5% 2025E Source: OECD, NBS, Morgan Stanley Research estimates Source: UN, Morgan Stanley Research estimates

Exhibit 20: Exhibit 21: China to attain high-income status as early as 2025 China’s long-term growth model 18,000 Base China: GNI per Capita (US$) Case (2025) 16,000 Bull Case 14,000 (2023) Bear 12,000 Case High-Income Threshold (2030) 10,000 2018 8,000 6,000 4,000 2,000 0 1990 1995 2000 2005 2010 2015 2020E 2025E 2030E Source: NBS, Morgan Stanley Research estimates Source: NBS, E= Morgan Stanley Research estimates

Will capex for Urbanization 2.0 lead to private investment in digital infrastructure with the liberaliza- tion of the telecommunications sector, which relies more on renewed debt problems? equity and corporate bond financing. l Better asset quality: Railway investment will focus more on A key market concern is whether increased capex demand for HSR in eastern China (where population density is high) and Urbanization 2.0 will cause a reaccelerated buildup in leverage, inter-city commuter rail to shorten daily commuting times, adding to financial stability risks. In our base case, we believe this risk which could generate stronger investment returns. We believe is manageable, considering: the NPL risks of such high-quality projects are relatively low. Moreover, wide adoption of AI and big data analytics could help l Less need for massive investment in Urbanization 2.0: improve banks' asset allocation and risk controls, ensuring China's urbanization over the past 40 years has built a strong healthier credit growth. foundation of infrastructure, reducing the need for massive investment. For instance, we estimate the combined capex Our China banks team estimates that the overall infrastructure needed for digital infrastructure, high-speed rail and the smart interest burden is still manageable at about 1.6% of GDP in 2019 (vs. grid – three key components of the smart supercity buildout – 1.59% for the US and 1.64% for Japan in 2018), and we believe policy- will be less than US$200bn per year in 2019-30, only about makers will continue to improve debt management through tight 10% of China's annual infrastructure FAI in the past five years. control of shadow bank financing. These factors combined echo our l More transparent funding: Over the past 2-3 years, policy- view that China will be able to stabilize its debt to GDP ratio in the makers have endeavored to improve the transparency of gov- next decade (see China: Blue Paper Revisit: Why we are still bullish ernment financing by gradually replacing local government on China, 14 November 2017). financing vehicle (LGFV) loans and shadow bank financing with local government bond issuance. Meanwhile, we expect more

24 BLUEPAPER

Bull-bearM scenarios Bull case: If China can accelerate Hukou and social security reforms Bear case: Slower-than-expected Hukou and land reform, wasted and complete the buildout of smart cities and city clusters at a faster investment and a lack of coordinated development in smart cities pace, the country's urbanization ratio could reach 78% by 2030 (vs. and city clusters, and larger-than-expected job losses due to automa- our base case of 75%). Against this backdrop, we would expect total tion and a weaker-than-expected external environment could factor productivity growth to remain elevated at 2.0% during dampen urbanization momentum. In this case, we would expect the 2019-30, contributing 40% of overall growth (compared with 30% urbanization ratio to reach 70% by 2030, only 10ppt higher than since the 2008 global financial crisis). Consequently, the country today. Productivity growth could soften at a faster pace, reaching could reach high-income status by as early as 2023, and gross 1.2% per year during 2019-30. Consequently, China would approach, national income per capita could reach US$21,800 by 2030. but narrowly miss, high-income status by 2030.

How could China mitigate the potential impact of automation on employment?

As technology and automation become entwined in almost every aspect of city life, a key market concern is that job creation could suffer as capital inputs with higher productivity could reduce the need for labor inputs, particularly in construction and lower value-added manufacturing sectors. We concur that such a scenario is a potential challenge for further urbanization, but believe it could be somewhat mitigated by China’s developing service sector, vocational training, and further reforms to the social security system. In the economic literature, automation typically affects labor demand in three channels:

1. Machines can completely replace labor in some sectors: History is not without episodes in which tech-enabled automation (such as mechanized looms in Britain in the nineteenth century) led to rapid and widespread labor replacement, causing, in some cases, unintended social issues for brief periods. 2. Higher labor demand in existing service sectors: A rise in capital productivity on the back of new technologies would give a boost to national income per capita. This would bring additional labor demand in existing service sectors that cannot be fully taken over by automation (such as barbers, fashion designers, health consultants, etc.). 3. Creation of new jobs alongside new technologies: The machines that replaced British textile workers also induced higher demand for engineers and maintenance workers, sales managers and delivery workers. In a more automated world, we also expect increased demand for jobs involving human interaction, creativity, and psychological counseling.

For most of the history of modernization, the latter two forces have combined to offset the first over the long run, as evident in the relatively stable share of household consumption in world aggregate output. In China, we believe the following factors could to some extent mitigate the risk of massive job losses and income disparity in the next decade: l Strong growth potential in services: In China, the service sector only accounts for 52% of GDP today, well below 68% in Japan and 80% in the US. As international experience suggests that the service sector's share of GDP increases as per capita income rises, we expect the ratio in China to reach 60% by 2030, helping absorb job losses from the manufacturing sector. Indeed, over the past five years, the number of people employed in the service sector increased by 63mn, more than three times greater than job losses of 17.8mn in the secondary sector. l Rapid development in vocational training business: Our education analyst, Sheng Zhong, believes that the overall vocational education and training market will grow 3x, to US$300bn in 2030. This will be supported by preferential government policies and binding subsidies. In our view, this could help displaced workers find jobs in other sectors (especially high-end, experience- driven services), reducing risk of frictional unemployment. l Further social security and welfare reforms:Policymakers have been easing Hukou restrictions and widening access to the social security system, which could provide some basic financial support to people undergoing frictional unemployment. That said, this combined with an aging population could add to the burden on the social security system. To this end, the government has initiated SOE asset transfers over the past two years to replenish the social insurance fund. We estimate that the further transfer of SOE equity to the social insurance fund (possibly up to 30% of total SOE equity vs. 10% now) and higher SOE dividend payout ratios (up to 50% vs. 33% now) could make up for the annual average social insurance deficit of Rmb577bn in 2014-19.

MORGAN STANLEY RESEARCH 25 M BLUEPAPER

Exhibit 22: Exhibit 23: China's services share of GDP to rise further with increased per capita State asset transfer to help address shortfall in social insurance bal- income ance amid aging population Share of services as % of GDP vs per capita income (USD) Rmb bn China's Demography and Social Insurance Solvency % True Balance of Social Security Fund 80% 300 34 100 Age Dependency Ratio-RS 36 -100 70% 38 -300 -500 40 60% -700 42 -900 44 50% China -1100 Germany -1300 46 40% Japan -1500 48 Korea 30% United States 2006 2008 2010 2012 2014 2016 2018 2020E 2022E 2024E 2026E 2028E 2030E 20% Source: CEIC, Morgan Stanley Research. E = Morgan Stanley Research estimates 0 10000 20000 30000 40000 50000 60000 Source: CEIC, Haver, Morgan Stanley Research. Data as of 1962-2018 for China and the US, 1965-2017 for Japan, 1972-2018 for Germany, and 1970-2018 for Korea.

26 M BLUEPAPER Initiative #1: City Clusters

Overview The combination of extensive high-speed railways and policy support has fostered the development of five key city clusters in China: Yangtze River Delta; Jing-Jin-Ji Area; Greater Bay Area; Mid-Yangtze River Area; and Chengdu-Chongqing Area. We believe city clusters are central to boosting productivity as they amplify the efficiencies of urban agglomeration while alleviating big-city problems. To ensure the success of city clusters, we expect additional policy reforms to enhance regional integration, more extensive transportation networks to foster a 'one-hour living circle' within clusters, and smart logistics systems to improve factor mobility.

Key forecasts The five key city clusters will account for about 75% of GDP growth and half of the urban population increase in 2019-30. The average population size per cluster should reach 120mn by 2030 (vs. 109mn today), close to Japan's current population of 127mn.

Why are city clusters important? included in the Yangtze River Delta (see Peixin Li, Chen Wang, Xueliang Zhang [2017] Did city cluster development help improve labor productivity in China? Volume 22, 2017, Journal of the Asia A shift in policy focus from regional rebalancing to city cluster Pacific Economy). development: As mentioned, China's top leadership put forward a new urbanization strategy in August with a focus on (1) strengthening coordination across local administrative boundaries within clusters What are the unique advantages of China's and (2) avoiding inefficiencies from limited demand for infrastructure city clusters? and services in lesser-populated, remote inland regions. This is in sharp contrast to past regional rebalancing initiatives (such as Despite the theoretical advantages of city clusters, in reality there 'Western Development' since 2000 and 'Northeast Revitalization' are only a few established ones in the world today. In China, since the since 2004), which aimed at reducing regional income gaps and 13th Five-Year Plan (2016-20) identified 19 city clusters, five have relieving the pressure of population inflows into developed coastal stood out – (1) the Yangtze River Delta; (2) the Jing-Jin-Ji Area; (3) the regions. Greater Bay Area; (4) the Mid-Yangtze River Area; and (5) the Chengdu-Chongqing Area ( Exhibit 24 ). We look at the advantages In our view, city cluster development will bolster productivity of these clusters in three respects: growth: According to the theory of agglomeration benefits, the bigger the city, the more productive it is likely to be, as it can better 1. Large populations: The average population size of the top match workers to jobs, facilitate the spread of ideas, shorten supply five clusters reached 109mn in 2018 – five times greater chains by gathering companies together, and generate more syner- than the New York Metropolitan Area, almost three times gies across different sectors. Recent OECD studies suggest that for higher than Greater Tokyo, and larger than most European each doubling in population size, the productivity level of a city countries ( Exhibit 25 ). This suggests significant potential increases 2-5%. In our view, the benefits of agglomeration can be for agglomeration benefits, in our view. Meanwhile, our amplified by well-integrated city clusters, which can also alleviate AlphaWise survey shows brighter job prospects and higher problems such as traffic congestion and high property prices that are wage growth in the five clusters than in other areas in often found in large cities. A study published in 2017 showed that in China (see Exhibit 26 and Exhibit 27 ), reflecting the China, counties enjoyed a 6% boost in productivity from being strong attraction of these areas to new migrants.

MORGAN STANLEY RESEARCH 27 M BLUEPAPER

2. Strong connectivity: China has the world's longest high- 3. Continued Hukou reforms: In China, possession of Hukou speed rail (HSR) network and the fastest operating speed (city resident permit) grants access to social security sys- (350km/h). This has cut travel time on major routes by tems, such as pensions, healthcare and education in that more than half. Meanwhile, the average distance per trip city. While Hukou control in megacities like Beijing and has been decreasing since 2013 ( Exhibit 29 ), suggesting Shanghai has remained tight, since April 2019 policymakers more frequent and short-distance travel within city clus- have completely removed Hukou restrictions for cities with ters. We believe the rail network enables trips of less than populations of 1-3mn (such as Zhuhai and Zhenjiang) and one hour within most city clusters. Such connectivity loosened restrictions in cities with populations of 3-5mn boosts productivity growth and enhances income conver- (such as Hefei, Nangtong and Huizhou). We believe con- gence among hub and satellite cities, as shown by the tinued reforms will create room for further population experiences of the Greater Bay Area and the Yangtze River inflows to smaller cities, particularly those in key city clus- Delta ( Exhibit 30 ). ters given brighter job prospects there.

Exhibit 24: China's five key city clusters

Jing-Jin-Ji Yangtze River Delta Greater Bay Area Mid Yangtze River Chengdu-Chongqing (京津冀城市群) (长三角城市群) (粤港澳大湾区城市群) (长江中游城市群) (成渝城市群)

Beijing Shanghai Jiaxing Shenzhen Wuhan Yueyang Chongqing Tianjin Suzhou Huzhou Guangzhou Changsha Yiyang Chengdu Shijiazhuang Nanjing Shaoxing Hong Kong Nanchang Changde Zigong Baoding Hangzhou Jinhua Foshan Huangshi Hengyang Luzhou Tangshan Ningbo Zhoushan Dongguan Ezhou Loudi Deyang Qinhuangdao Hefei Taizhou (ZJ) Macau Huanggang Jiujiang Mianyang Langfang Wuxi Wuhu Huizhou Xiaogan Jingdezhen Suining Cangzhou Changzhou Maanshan Zhongshan Xianning Yingtan Neijiang Chengde Nantong Tongling Jiangmen Xiangyang Xinyu Leshan Zhangjiakou Yancheng Anqing Zhuhai Yichang Yichun Nanchong Yangzhou Chuzhou Zhaoqing Jingzhou Pingxiang Meishan Zhenjiang Chizhou Jingmen Shangrao Yibin Taizhou (JS) Xuancheng Zhuzhou Fuzhou Guangan Xiangtan Jian Dazhou Yaan Ziyang

Source: NDRC, Morgan Stanley Research. Cities in pink are cities currently with urban populations close to or above 8mn, and cities in yellow are emerging mega cities in which we expect the urban population to exceed 8mn by 2030.

28 M BLUEPAPER

Exhibit 25: Exhibit 26: China's top five city clusters are larger than the Greater Tokyo Area, the Brighter job prospects... New York Metropolitan Area and most EU countries Perception of Job Prospects in the City: 2019 vs. 2017 Population Size (people mn) 2017 Survey 2019 Survey

76 74 74 Avg of China's Five Key Clusters 109 72 73 72 70 68 66 Germany 82 60 53 53 U.K. 67 France 65 Italy 59 Spain 46

Greater Tokyo Area 44 Rating 'Optimistic')(% New York Metropolitan 20 Mid Yangtze Yangtze River Jing-Jin-Ji Guangdong Chengdu Other 0 20 40 60 80 100 120 River Delta Bay Area Chongqing Source: CEIC, Haver, Morgan Stanley Research. Data as of 2018 Source: AlphaWise, Morgan Stanley Research

Exhibit 27: Exhibit 28: ...and higher wage growth in China's five key city clusters China's high-speed rail is the fastest in the world 2019 Survey: Income Growth Perception Regular Operating Speed, Km/h by City Cluster Current vs. 12 months ago Next 12 months vs. Current China 350 56 51 Germany 320 48 46 Spain 320 39 37 37 39 36 35 France 320 31 Japan 320 26 Italy 300 Taiwan 300 Korea 300 Russia 250 Yangtze River Chengdu Jing-Jin-Ji Other Guangdong Mid Yangtze US 240 Delta Chongqing Bay Area River Net Score (% Up MinusDown) Net (% % Score 200 250 300 350 Source: AlphaWise, Morgan Stanley Research Source: CIA, Morgan Stanley Research. Data as of 2018

Exhibit 29: Exhibit 30: More frequent and short-distance travel within city clusters Greater infrastructure connectivity has led to robust income conver- Average Railway Travel Distance Per Person, Km gence 532 532 540 527 524 523 13% 518

517 520 516 503 12% 2018 500 488 y = -0.00071x + 0.124 11% R² = 0.50 480 472 10% 460 447 436 9% 440 419 420 8%

400 7% 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Real CAGR during 2010- 6% Source: CEIC, Morgan Stanley Research 10 30 50 70 90 GDP per capita in 2009, RMB

Source: CEIC, Morgan Stanley Research

MORGAN STANLEY RESEARCH 29 M BLUEPAPER

Fact sheet on China's top five city clusters

Economic structure: The economic output of the top five city clusters was US$7.6trn in 2018, accounting for 54% of China’s GDP. In the Greater Bay Area, Jing-Jin-Ji and the Yangtze River Delta, the share of the service sector in GDP is higher than the national average, while the Mid-Yangtze River Area and the Chengdu-Chongqing Area are still more focused on the industrial sector. Pillar industries in city clusters: l Yangtze River Delta – finance, manufacturing (heavy machinery and auto), IT l Jing-Jin-Ji – manufacturing (general equipment, steel, aviation), cultural l Greater Bay Area – finance, tech innovation, manufacturing (home appliances, electronics), services (professional and entertainment) l Mid-Yangtze River – heavy industry, manufacturing (autos & parts, equipment, consumer goods) l Chengdu-Chongqing – manufacturing (consumer electronics, equipment, food & beverage), tourism

Population growth: In 2010-18, population growth was strongest in the Greater Bay Area (1.4% CAGR), followed by Jing-Jin-Ji (1.1%), Yangtze (0.9%), Cheng-Yu and Mid-Yangtze (0.6% for both), and all were above the national average of 0.5%. However, population flows within hub and satellite cities show a mixed picture. In the Greater Bay Area, Cheng-Yu and the Mid-Yangtze River Area, hub cities accounted for more population inflows in 2015-18 than in 2011-14, and hub cities in the Yangtze River Delta have still maintained their attractiveness, taking up 57% of new inflows in 2011-18 as tier-2 cities, such as Hangzhou and Nanjing, have taken over the role of Shanghai (which has a strict Hukou policy) in attracting new migrants. In contrast, satellite cities in Jing-Jin-Ji are more attractive than its hub cities, as Beijing has targeted relocating 'non-capital functions' to nearby cities.

Exhibit 31: Overview of top five city clusters, 2018

Yangtze River Mid-Yangtze Chengdu- Five Jing-Jin-Ji Greater Bay Area National* Delta River Chongqing Clusters*

Area (000' sqkm) 213 183 56 343 240 1,035 9,597 Population (mn) 154 89 71 127 100 541 1,403 Population Density (ppl/sqkm) 723 485 1,268 369 418 522 146 GDP (USD bn) 2,672 1,150 1,625 1,245 860 7,552 13,880 GDP per capita (USD) 17,351 12,932 22,913 9,817 8,590 13,965 9,891 • Finance • Finance • Shipping • Heavy Industry • Capital • Tech & Innovation • Internet Service • Capital • Capital Equipment • Professional Service Major Industries • Shipping Equipment Equipment • Tourism -- • Shipping • Auto & Parts • Cultural Industry • Consumer • Food & • Entertainment • Heavy Machinery • Aerospace Goods Beverage • Home Appliance

Source: CEIC, Haver Analytics, local government websites

30 M BLUEPAPER

Exhibit 32: Exhibit 33: GDP structure, by city cluster, 2018 Population growth of top five city clusters GDP Structure, % of GDP Tertiary Secondary Primary (2018) Greater Bay Area 1.4% 100% Jing-Jin-Ji 1.1% 80% Yangtze 0.9% 60% Major Clusters 0.8% 40% Cheng-Yu 0.6% 20% Mid-Yangtze 0.6% Annual Population 0% growth, 2011-2018 Total 0.5% Yu Ji Bay

Mid- 0.0% 0.5% 1.0% 1.5% Area* Cheng- Greater Yangtze Yangtze Jing-Jin- National

Average^ Source: CEIC, Morgan Stanley Research

Source: CEIC, Morgan Stanley Research. Note: *2016 data for Hong Kong and Macau. ^National average excludes Hong Kong and Macau

Exhibit 34: Exhibit 35: Percentage of population inflows taken by hub cities Contribution to population increase, by city in the Yangtze River Delta 2011-2014 2015-2018 100% Contribution to 90% population increase 120% % of population inflow taken by 113% 80% hub cities Satellite Cities 100% 70% 81% 83% 83% Suzhou 80% 60% 65% Nanjing 57% 57% 61% 50% 60% Ningbo 40% 42% 38% Hangzhou 40% 30% Shanghai 20% 20% Hefei 10% 0% Jing-Jin-Ji Yangtze Greater Bay Cheng-Yu Mid-Yangtze 0% Area 2011-2014 2015-2018 Source: CEIC, Morgan Stanley Research Source: CEIC, Morgan Stanley Research

MORGAN STANLEY RESEARCH 31 M BLUEPAPER How can city clusters be promoted further? point, the blueprint for the Shenzhen Demonstration Pilot Zone (August 2019) included several institutional frameworks for regional coordination, such as integrated planning for City clusters are more than just large populations and integrated Shenzhen-Dongguan-Huizhou and a special coordination com- infrastructure. A common concern is that the blind development of mittee for the Shenzhen-Shantou Cooperation Zone. city clusters may lead to debt problems and wasted resources, rather than productivity growth. In our view, the keys to reaping the benefits 2. More advanced railway network of city clusters are: (1) freer factor/labor mobility, and (2) better coordination among cities. In this context, we believe further reforms to Although China already has the world's longest and fastest high- ensure regional integration could be supportive, as would more speed rail system, it's not finished yet. Our industrials analyst, Kevin advanced railway networks to enhance inter-city and intra-city com- Luo, expects the length of the HSR system to reach 65,000km by muting, and smarter logistics systems: 2030, up from 30,000km now. Meanwhile, the HSR mix will likely improve as Kevin estimates that half of HSR lines to be completed by 1. Further policy reforms to enhance regional 2030 will be in eastern China, up from 33% in the past five years. integration The team also expects the length of the inter-city commuter rail net- A key challenge to city cluster development is the segmented social work to expand to 17,000km by 2030 from just 2,000km today, security system and independent government planning across cities, while the length of metro lines will reach 15,000km from 5,800km which has not only deterred labor mobility but also led to duplicate in 2018. This will further enhance inter-city connections and sub- construction and overcapacity. We thus believe future policy reforms urban commuting, bringing more population flows to suburbs and should focus on: satellite cities with lower housing prices and more living space, easing congestion in city centers (see 3a. Transportation for more details). l Standardizing social security systems across cities: While policymakers have gradually reduced Hukou restrictions for 3. Smarter logistics system smaller cities, access to city-specific pension, medical insurance, and education systems still requires lengthy administrative pro- A smarter logistics system enabled by IoT, big data and a better trans- cedures for inter-city migrants, which hinders labor mobility. portation network would help lower logistics costs, shorten delivery This suggests a need for integrated social security systems (e.g., times and address last-mile delivery, facilitating factor mobility in city in the form of a universal residence passport) in city clusters, clusters. For instance, big data analytics could help predict customer with the current degree of integration still quite limited (for demand and automatically arrange for product to be shipped in example, residents of Shenzhen, Dongguan and Huizhou can advance to the warehouses/stores that are closest to customers, and claim medical insurance in 18 hospitals across the three cities). automated sorting lines and industrial robots could make warehouse l Establishing a public data-sharing system within clusters: sorting and packing more efficient. Meanwhile, delivery times could To facilitate an integrated social welfare system, a transparent be reduced, with trains gradually replacing trucks in long-haul trans- public data-sharing system is necessary. The shared database of portation. With the development of technology, the adoption of social benefits, medical records, and education progress would unmanned drones/trucks would not only reduce efficiency losses by make it easy for people to enjoy the same public services in any delivering goods around the clock, but also provide low-cost delivery city within a cluster. Meanwhile, an integrated public database options to easily reach low-density suburban areas. would enable one-stop information provision in a single government mobile app, such as traffic conditions, public events, In this context, transportation analyst Qianlei Fan believes average shopping promotions, and ticket discounts at attractions in dif- delivery times could be shortened to 24 hours nationwide by 2030 ferent cities, helping people plan trips in the cluster. (vs. 2-4 days today) and 12 hours within each city cluster (vs. 24 hours l Forming a more coordinated regional development in the Yangtze River Delta today). Meanwhile, logistics costs as a per- strategy: To maximize synergies across cities, local govern- centage of GDP could come down to 10% by 2030 from 15% today, ments should work together to ensure more proper industry and express volumes could reach 300bn deliveries per year as com- distribution in a region, based on cities' respective strengths, so pared to 50bn today (see 2b. Logistics for more details). as to reduce repetitive production and competition while increasing cooperation across the supply chain. As a case in

32 M BLUEPAPER Case study: Development of the Yangtze Outlook: Increasing economic importance River Delta of five key clusters

The Yangtze River Delta – comprising Shanghai, Zhejiang, Jiangsu, We expect China's top five city clusters to account for 75% of China's and Anhui – has led city cluster development in China. It has the real GDP growth in 2019-30 (vs. 65% in 2014-18), with annual real highest GDP and largest population of the five key city clusters, GDP and total factor productivity growth sustaining at around 6.0% launched high-speed rail as early as 2010, and possesses competitive and 2.0%, respectively – higher than the national averages of 4.6% advantages in diversified industries. It contains China's key financial and 1.6%. GDP per capita in these clusters, which was US$13,000 in and business center (Shanghai), the hub of internet giants 2018 and has already exceeded the high income threshold of (Hangzhou), the nation's key manufacturing hub for autos and heavy US$12,300 (as of 2018), will almost double to US$25,000 by 2030, machinery, and a strong shipping industry. we project. Meanwhile, these clusters will likely absorb 50% of the incremental urban population, with the total urbanization ratio Over the past two years, policymakers have intensified the push to picking up from 67% currently to 80%, similar to the level in the US enhance the region's coordinated development. In 2018, the local and Japan in the early 2000s. We expect the average population size governments of Shanghai and three provinces in the region estab- of these five city clusters to reach around 120mn by 2030 (vs. 109mn lished the Yangtze River Delta Regional Cooperation Office, and today), close to Japan's total population size of 127mn today. issued a three-year action plan (2018-20) for the region's integration. This year, Beijing further promoted the region's development as a national strategy in the 2019 Government Work report, and the CPC Exhibit 36: Central Committee reviewed the central government's blueprint for Productivity in the five key clusters to hold up... the region in May, which should be published relatively soon. TFP Growth, CAGR % 2.4% 2.3% Five key City Clusters Nationwide

In our view, policymakers will likely make more efforts on the fol- 2.2% 2.0% lowing aspects to enhance integration within the region: 2.0% 1.9%

1.8% 1.7% l Unified social benefits system, which would enable residents 1.6%

to enjoy the same pension and healthcare services in any city 1.4% within the cluster. As an initial step, the Yangtze River Delta 1.2% development plan mentions the promotion of easy pension 1.0% transfers within the cluster. 2014-18 2019-30E l Integrated public database to increase information transpar- Source: NBS, Morgan Stanley Research estimates ency. Today, the region has set up an integrated government website covering 14 cities, which provides corporate registra- Exhibit 37: tion services, pre-booking for marriage registration, and guide- ...helping double average income despite a high starting point Per Capita Income, USD lines for various cross-city administrative work, such as pension 30,000 transfers and medical insurance. Five key City Clusters 25,500 l 25,000 Standardized transportation network, which will facilitate Nationwide

travel within the cluster. The region plans to unify the metro 20,000 18,292

mobile payment system in nine cities by end of this year, so 15,000 13,488 that people can use the same QR code as e-tickets to take 9,452 10,000 metros in different cities. l Coordinated industrial development to reduce repetitive pro- 5,000 duction and increase synergies. This could be facilitated by 0 2018 2030E better industrial information sharing within the cluster. Source: NBS, Morgan Stanley Research estimates l Simplified administrative procedures for companies doing business in different cities in the region.

MORGAN STANLEY RESEARCH 33 M BLUEPAPER Initiative #2: Smart Cities

Overview Data-driven smart cities can be optimized to accommodate larger populations. Specifically, we expect smart cities to enable faster commuting, provide a safer living environment by reducing crime and traffic accidents, lower pollution through the use of EVs and green energy sources, and make cities more livable via smart household appliances and improved education and healthcare. We believe the pace of smart city growth in China will be faster than in many other countries, given government efforts to boost the development of fiber networks, 5G, big data, AI, and edge computing.

Key forecasts We expect the smart city buildout to boost cities' capacity to accommodate more population, lifting China's number of megacities with populations similar to or larger than New York City (8mn) to 23 by 2030 vs. 9 today.

What benefits can a smart city provide? That said, we believe the development of smart cities can effectively improve city management and boost the capacity of large cities to accommodate more people. In other words, large cities A key concern about urbanization is whether the growth of cities today have the potential to be even larger over the next decade. will reach a limit. This concern has increased in view of negative pop- Specifically, we believe a well-developed smart city powered by ulation growth in Beijing and Shanghai over the past two years. advanced technologies (such as 5G and IoT, cloud computing, AI and According to a 2011 McKinsey report, the key hurdles to develop- big data analysis) could enable faster commutes, safer cities with ment are a lack of skillful planning and management to handle a reduced crime rates and traffic accidents, and better living environ- growing population. This could foster various big-city problems, such ments with smarter household living, easier access to public as congestion, higher crime rates, severe competition for social resources, and less pollution in the coming decades. resources, and pollution, outweighing scale benefits and leading to a deterioration in quality of life.

Exhibit 38: China's future smart cities have the potential to be faster, safer, greener, and more livable

Source: Morgan Stanley Research

34 M BLUEPAPER 1. Faster commutes first 5G-based smart highway project to support self-driving cars. In this context, our automobile team forecasts that 20% Traffic congestion has long been a headache for city dwellers, and the of the passenger vehicles sold in 2030 will feature L4 or L5 degree of congestion tends to increase with the size of the urban pop- levels of autonomous driving vs. 0% today. ulation ( Exhibit 39 ). Amap’s 2018 Traffic Analysis report shows that the average weekday commute in China’s top ten cities (by urban Exhibit 39: population size) is 79 minutes, with 45% of the time spent in conges- Traffic congestion tends to increase with city size tion, and we estimate the nationwide average commute time is about 55% one hour. That said, this condition could significantly improve in the 50% Beijing next decade through smarter traffic control systems, shared Guangzhou Chongqing 45% Shenzhen Shanghai mobility, and auto driving. Moreover, a reduced need for people to Chengdu Dongguan 40% Wuhan Tianjin drive suggests they will have more time for work and entertainment, Suzhou increasing productivity in the economy. 35% l 30% Smarter traffic control: A smart traffic control system could TimeSpenton Congestion Urban Population (Mn) as a % ofa as % TimeTotalCommuting help optimize traffic flows by adjusting traffic lights in intersec- 25% 0 2 4 6 8 10 12 14 16 18 20 22 24 tions, providing route suggestions, and helping to locate Source: Amap 2018 Traffic Analysis Report, NBS, Morgan Stanley Research parking spaces in a timely fashion. The rapid development of 5G and IoT is making this possible. Hangzhou is the first pilot Exhibit 40: city in China to adopt a smart traffic control system – its 'City Shared mobility could dampen willingness to buy new cars Brain' project conducts real-time monitoring of traffic flows % Intending to buy a private car irrespective of taxi-hailing app with camera systems and sensors. AI can then optimize traffic 2017 survey 2019 survey signals at over 100 intersections to reduce congestion and pri- 70 66 63 oritize ambulances and fire engines. The City Brain system can 61 also detect traffic accidents immediately and improve the efficiency of traffic police. In Shanghai, a smart parking network launched by Huawei will enable drivers to locate available parking lots nearby, reducing unnecessary driving. l Shared mobility: Online taxi-hailing platforms should help Tier 1-2 Tier 3 or lower enhance asset efficiency and reduce the number of cars on the Source: AlphaWise, Morgan Stanley Research road, alleviating traffic pressure. Our AlphaWise survey in 2Q19 showed that the percentage of surveyed households intending Exhibit 41: to buy a private car irrespective of taxi-hailing apps has Six levels of autonomous driving declined in both large and smaller cities as compared to the 2017 survey ( Exhibit 40 ). AI and smart tariff systems could also better match customers and drivers, reducing wait times for cars. l Autonomous vehicles: The low latency of 5G means IoT-based autonomous driving is not a distant dream. Industry body SAE sets six levels of autonomous driving, from no automation (Level 0) to full automation (Level 5) ( Exhibit 41 ). In our view, China will likely adopt autonomous vehicles at a faster pace than other countries owing to government support, rapid infrastructure buildup, and the higher possibility of commercialization given the significant market size. As a case in point, in February 2019, Hubei Province started work on the country's Source: Shutterstock

MORGAN STANLEY RESEARCH 35 M BLUEPAPER 2. Safer society 4. More livable environment

Crime and traffic accidents are two of the main safety concerns in Another challenge of big cities is quality of life, including the poten- cities. Today, the development of cashless payment in China has tial for reduced personal time as a consequence of busy work sched- already helped reduce cases of robbery and theft. As we have fully ules, strong competition for high-quality public resources, and rising entered the digital age, cybersecurity is becoming increasingly impor- levels of environmental pollution. We believe smart cities can help tant, and AI and big data analysis could help identify wire fraud by address these issues through: detecting unusual transactions and login locations. Meanwhile, massive adoption of smart cameras with video analytics can promptly l Smarter home appliances: We are already able to purchase verify people’s identities and locate suspects, and AI and big data ana- goods and services online and remotely control household lytics can help predict locations of possible crimes. We also expect appliances via mobile apps. The wider usage of big data analysis there to be fewer traffic accidents in smart cities. While more intelli- and IoT could provide more customized services. For instance, a gent traffic control systems should help optimize traffic flows, the smart housing management system could monitor air quality wide adoption of interconnected driverless cars based on 5G and IoT and dust and decide when an air purifier or robot vacuum could make timely adjustments based on real-time road conditions cleaner should be activated; a smart refrigerator could auto- and reduce human error. matically throw away expired food, order staple foods (such as eggs and milk) online, and provide cooking materials to a smart 3. Greener life kitchen; and a smart washing machine would sort clothes into different batches by fabric and decide on the most suitable We believe the wider adoption of electric vehicles and green energy washing mode. Moreover, the future development of autono- sources can effectively reduce air pollution. To this end, our automo- mous delivery associated with e-commerce could further bile team forecasts that one-third of passenger vehicles sold in China enhance the online shopping experience. in 2030 will be electric, vs. only 4% in 2018. Meanwhile, given the l Smarter healthcare system: While AI and big data analytics Chinese government’s commitment to combating global climate could better match healthcare resources with patients, create change, our utility team expects the share of non-fossil fuels (hydro, cloud hospitals, and improve the accuracy of diagnosis, 5G and nuclear, wind and solar) in total power generation to increase to 40% IoT could also enable remote surgery, breaking geographic con- in 2030 from 30% in 2018. The team also believes that investment straints on high-quality medical resources. Cities including in the smart grid could increase 2.64x, to US$80bn during 2021-30, Ningbo and Guangzhou have started to run online medical plat- to integrate distributed renewable energy and electric vehicles and forms for citizens, to perform pre-diagnosis and provide access ensure the safety and reliability of the power system. to suitable healthcare services. l Smarter education system: AI and big data analysis can better track a student's mastery of subjects and provide more personalized assignments as well as suitable interactive online courses.

36 M BLUEPAPER Enablers of smart cities in China urbanization, construction and management that integrates advanced technologies, information resources and business application systems. MOHURD pointed out that the construction of smart In our view, smart cities could be developed more quickly in China cities is an important way to realize the State Council’s guidance of than in many other countries owing to government support. Since innovation-driven development and urbanization in China. policymakers first raised the concept of smart cities in 2012, more Qualifying cities were encouraged to take advantage of the opportu- than 500 cities have initiated city planning toward this end. nity and apply for the pilot project to promote industrial transforma- Meanwhile, strong policy support has been provided to build the key tion and development and improve the level of city management and elements of smart cities, including an extensive fiber network, 5G, big services. data, AI, and edge computing. Starting from 2013, MOHURD has announced three tranches of trial Policy support smart cities, comprising 290 cities and districts. In 2014, eight ministries together published their guidance on smart city development, In 2012, the Ministry of Housing and Urban-Rural Development of setting a goal to complete the construction of several characteristic China (MOHURD) released a notice that it was launching the Smart smart cities by 2020, which should be the starting point for future City Pilot Project. MOHURD defined smart cities as a new model of smart city clusters.

Exhibit 42: Key trial smart cities

Source: MOHURD, Morgan Stanley Research

MORGAN STANLEY RESEARCH 37 M BLUEPAPER

Exhibit 43: Key policies issued by the central government

Source: Ministry of Housing and Urban-Rural Development (MOHURD), National Development and Reform Commission (NDRC), Ministry of Land and Resources (MLR), Ministry of Transport (MOT), Ministry of Industry and Information Technology (MIIT), Ministry of Natural Resources (MNR), Ministry of Science and Technology (MST), Ministry of Public Security (MPS), Ministry of Finance (MOF); eight ministries include NDRC, MIIT, MST, MPS, MOF, MLR, MOHURD, MOT

38 M BLUEPAPER Infrastructure: Fiber and 5G Cloud

China is one of the most fiberized countries globally, especially con- China has seen a ten-fold expansion of the cloud market over the past sidering its large geographical size and relatively low population den- decade. According to IDC, China was the second-largest IaaS market sity as compared with countries such as Japan and Korea. As of June at the end of 2018. And the total cloud service market 2019, according to MIIT, total optical fiber length reached 45mn kilo- (IaaS+PaaS+SaaS) is expected reach US$27.5bn by 2022. The fast meters. Fiber-to-the-home/office (FTTH/O) subs reached 400mn, ramp-up of cloud could provide solid support for the rollout of smart representing over 90% of total broadband subs in China. Many smart cities as: home applications, such as connected home appliances, surveillance and entertainment, require extensive FTTH penetration. In China, l It provides a platform to connect all data generated from the almost all of the mobile base transceiver stations (BTS) are con- digitization and informatization of a city's functions, including nected with fiber for transmission, which is essential to providing utilities, transportation, healthcare, and education. The elimina- 4G/5G services. tion of isolated silos of information could translate into better efficiency in the daily operation of the city 5G will offer unprecedented bandwidth, low latency, fast mobility, l It provides a platform for better management of the massive and high capacity, which will support a set of brand new applications number of end-devices connected to the network with the on top of a better experience using traditional smartphone connec- emergence of various new use cases tions. Based on the network capability requirement, 5G applications l Virtualization technology on the cloud platform indicates a can be classified into three key categories: better arrangement of network resources and real-time adjustment according to different situations and emergencies. l Enhanced mobile broadband (eMBB): enhanced indoor and out- door broadband, enterprise collaboration, augmented and vir- Cybersecurity tual reality l Ultra reliable low latency communications (URLLC): autono- The security of cloud platforms, IT systems and sensitive information mous vehicles, smart grids, remote patient monitoring and tele- from the government and public infrastructure has become a signifi- health, industrial automation cant concern for smart cities. Even small leakages or short shut- l Massive machine type communications (mMTC): IoT, asset downs of key infrastructure have the potential to cause major losses tracking, smart agriculture, smart cities, energy monitoring, for a city. On 23 December 2015, the power supply systems in three smart homes, remote monitoring areas of the Ukraine were attacked by malicious software and shut down for six hours, affecting over 1.4mn people. Such attacks have China is one of the leading 5G pioneers globally. More important, become more frequent in the past few years and increasingly target while other markets such as the US focus more on consumer applica- key facilities in major cities. According to a survey done by CNCERT tions (e.g., fixed wireless), the Chinese government and operators in 2018, malicious attacks in Beijing and Shanghai accounted for have always been focused on industrial IoT applications for enter- about 10% of total attacks in China. And provinces with more big prise and government segments. cities, such as Jiangsu and Guangdong, tend be at greater risk of attack than other provinces. Therefore, we believe more advanced cyberse- Edge computing refers to infrastructure that enables data pro- curity technology and comprehensive solutions for digitalized infra- cessing as close to the source as possible, which allows for faster pro- structure will help guarantee the smooth functioning of smart cities. cessing of data, reducing latency and improving the customer experience. This type of computing will require micro-data centers China has underspent on cybersecurity compared with other coun- close to 5G BTS. The edge computing discussion is still in the very tries in the past. However, it is expected to be the sector with the early stages, as potential applications, like autonomous driving, are fastest growth in terms of IT spending in the next decade. According several years away. We also note that micro-data center investment to IDC, spending on cybersecurity in China because of government could be shared by telcos (network), data center operators (physical encouragement will be the main growth driver. The Chinese govern- environment) and information service providers (computing ment has launched several laws and regulations, including the first capacity), and future development will be connected with specific China Cybersecurity Law, Critical Information Infrastructure Security applications. Protection Regulation, Graded Protection 2.0 Policy, etc. We believe

MORGAN STANLEY RESEARCH 39 M BLUEPAPER the strong push from those policies and detailed standards should We believe smart city development could effectively enhance the help boost growth in the smart security supply chain and accelerate capacity of the cities of tomorrow. We expect the combination of smart city development. smarter traffic control systems, shared mobility, and auto driving could significantly reduce congestion, shortening the average commute time from home to office to merely 15 minutes by 2030. The Outlook: More megacities to emerge in deployment of cybersecurity, AI, and automated vehicle technolo- the next decade gies should help make streets and roads safer than ever. Moreover, digitalization and advanced technology can enhance connectivity in every aspect of urban life. Lillian Lou, our China consumer analyst, In the past, Hukou restrictions, in part a response to rising urban believes that average consumer IoT devices per household could problems in larger cities, have restricted the growth of China's larger reach 7 units by 2030 (vs. 1 today), which could free people from cities. Contrary to common perceptions, China’s large cities are too cleaning, cooking, doing laundry, and grocery shopping in favor of small to be consistent with Zipf’s law, which posits that the growth work and entertainment. rate of cities should be independent of their size because the efficiency gains from agglomeration will offset related costs, and as a In this context, we expect the 50 largest cities in China to expand 3% result, for a group of cities, if one plots the log of population and log per year (measured by urban population) towards 2030, with the of corresponding ranking (in terms of population size) in a scatter remaining cities growing by 2.5%, reversing the trend in 2010-18, chart, the graph will mimic a straight line, often with a slope of -1 when the top 50 cities grew by 2.4%, compared with 3.1% in other ( Exhibit 44 ). For China in 2018, as in Exhibit 45 , more large cities cities. Consequently, the number of Chinese cities with populations deviated away from the fitted line, whose slope is also much steeper exceeding 8mn will likely jump to 23 in 2030 from 9 today, while than -1, suggesting that the growth rate of China's cities has been neg- those with more than 5mn will also pick up significantly, to 50 from atively correlated with their size. 33.

Exhibit 44: Exhibit 45: Zipf's law for the US Zipf's law for China Size/Rank Distribution of Top 250 US Metropolitan Areas Size/Rank Distrubution of Top 250 Chinese Cities (2018) 6 6

5 5

4 4 y = -1.00x + 10.91 R² = 0.98 3 3 y = -1.52x + 5.91 2 2 R² = 0.95 Natural Log of Rank Natural Natural Log of Rank 1 1

0 0 5 6 7 8 9 10 011223344 Natural Log of Population Natural Log of Urban Population

Source: Haver, Morgan Stanley Research Source: Government website, Morgan Stanley Research

Exhibit 46: Exhibit 47: Continued urbanization... …will create larger cities Urbanization Ratio, % Projected Number of Cities Above Certain Size

100% China 50 90% Japan 2018 2030E 80% 2030E: US 70% 75% 40 60% 2018: 50% 59.6% 30 40% 30% 20 20% 10% 0% 10

1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2018 0 2025E >8m >5m Source: UN, NBS, Morgan Stanley Research estimates Source: Local government websites, Morgan Stanley Research estimates

40 M BLUEPAPER Key smart city applications divided into three groups – populations of below 15mn, between 15mn and 20mn, and above 20mn – and the crime index of these cities steps up with the population. Apart from public security con- Smart security cerns, risks from traffic, fire and other infrastructure incidents increase significantly. Therefore, improving a city's security manage- Increased security concerns have become a major obstacle to the ment through technology is one of the major steps for further expan- expansion of cities. According to the National Bureau of Statistics, sion of cities. total crime cases filed by police departments almost doubled from 2001 to 2017, along with the accelerating rampup of urbanization in Policy support from government could be the main growth China, from 38% to 58%. This situation is even more obvious at the driver. For public security, mainly surveillance and traffic monitoring city level, as highly populated cities tend to have a lower sense of systems, the government has been rolling out several nationwide safety. Based on a study by Numbeo, major cities in China can be projects since 2000.

Exhibit 48: Exhibit 49: Total number of crime cases and urbanization rate in China (2001-17) Comparison of the crime index and populations of different cities in

Crime cases (000) Urbanization rate (%) 8,000 60% China

7,000

55% 6,000

5,000 50%

4,000

45% 3,000

2,000 40%

1,000

- 35% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: National Bureau of Statistics, Morgan Stanley Research

Exhibit 50: Major surveillance projects launched by government since 2000

Source: Numbeo, Morgan Stanley Research. Data as of 2018.

Exhibit 51: Cybersecurity policies issued in recent years

Date Policy Authority

Feb-14 The Central Leading Group for Cyberspace Affairs State Council

Government Guidance on Enhancing Telecom and Aug-14 Ministry of Industry and Information Technology Source: Morgan Stanley Research Internet Security Government Opinion on Enhancing Military IT 14-Oct Central Military Commission Security

Nov-16 Cybersecurity Law National People's Congress

Dec-16 13th Five-Year Plan for National IT Industry National Development and Reform Commission

Opinions on Promoting the Sound and Orderly Jan-17 State Council Development of the Internet Guiding Opinions on Promoting Capital Markets in Mar-18 The Central Leading Group for Cyberspace Affairs Serving the Cyberspace Power Building

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 41 M BLUEPAPER How will smart security be built? We expect demand from the government to grow fastest, with the installation base rising at a 20% CAGR during 2018-30. Additional New smart security use cases keep emerging as the requirements of cameras will be installed to deter crime. Traffic monitoring systems security continue to evolve. With the expansion of city sizes, higher combine surveillance cameras from train stations, airports, and main populations, increasing traffic volumes, and greater residential inten- roads, and traffic data is analyzed in real time at the back-end server sity, demand for security management will increase. with AI enabled. Such systems can better manage traffic and avoid traffic jams. Other than human and vehicle identification, these AI In the current stage, the use cases for smart security are mainly in the cameras are capable of floating object detection, sewage disposal following areas: detection, and wild animal detection. l Crime investigation We expect the professional surveillance camera installation base to l Traffic monitoring reach 897mn by 2030, at a 12% CAGR during 2018-30. The number l Residential area surveillance of surveillance cameras will increase to 62 per 100 people by 2030 l Industrial and public infrastructure security from 16 per 100 people in 2018, we estimate. This would be significantly higher than levels in the rest of the world, which we expect The adoption of new technologies, especially AI, and the maturing to reach 17 cameras per 100 people by 2030. We expect penetration domestic supply chain also aid in the accelerating development of of AI-embedded cameras to reach 14% by 2030 from less than 1% in smart security in China. 2019. Growth will be supported by demand from the government (such as the Xueliang Project) and commercial applications, such as l Edge computing: facial recognition, object detection and video unmanned supermarkets. structuralization l Cloud platform: big data analysis, database searching, and We also expect the consumer surveillance camera market to grow at smart analysis a 30% CAGR during 2018-30. According to IDC, 9.7mn security cameras were shipped to consumers in China in 2018, mainly for home How large could the surveillance market be? security. We believe more households have demand for these cameras, which are able to recognize strange faces, detect abnormal With AI embedded, surveillance cameras are becoming an important behaviors such as falling down, and to monitor and record the impor- tool to improve efficiency and productivity in smart cities. We esti- tant moments of babies or pets. IHS estimates that the number of mate the professional surveillance camera market will grow at an 8% households with cameras will increase to 67mn by 2030 from just CAGR during 2018-30, fuelled by new installations until 2025 and 8.3mn in 2018, and camera installations per household will increase replacement demand thereafter. to 5.9 by 2030 from 2.3 in 2018.

Exhibit 52: Exhibit 53: China: Surveillance camera installation base China: Surveillance camera shipments 1,000 200 mn mn

800 160

600 120

80 400

40 200

0 0

Net Add Replacement

Source: IHS, Morgan Stanley Research estimates Source: IHS, Morgan Stanley Research estimates

42 M BLUEPAPER

Exhibit 54: Exhibit 55: China: Surveillance camera ASP and surveillance equipment market Number of surveillance cameras per 100 people size 80 China 25,000 60 USD mn USD Global 60 20,000 50

15,000 40 40 10,000

30 20 5,000

0 20 0 2018 2030E Source: IHS, Morgan Stanley Research estimates Revenue ASP Source: IHS, Morgan Stanley Research estimates

How can smart security enable the development of smart cities? primary objective of adopting V2X is to reduce accidents, improve efficiency and save resources. V2X contains three key aspects: vehi- There have already been quite a few successful cases in Shanghai and cle-to-infrastructure (V2I), vehicle-to-vehicle (V2V) and vehicle-to- Hangzhou: network (V2N). While both V2I and V2V will likely play the key role in providing connectivity to cars, we believe that in a world where l Shanghai: AI surveillance cameras, working as electronic police, there is full car autonomy, mobile networks will need to play a crucial have captured hundreds of motorists running red lights. The role in the ecosystem in the following areas: whole system is an integrated application, including video analysis, motion tracking, and facial recognition. Surveillance cam- l Human safety. In our view, vehicle connectivity to mobile net- eras are also applied to monitoring garbage sorting. works can address crashes that cannot otherwise be prevented l Hangzhou: It would normally take roughly 30 days and 1,500 by current technology (using camera and sensors) or vehicle-to- police officers to search 250 hours of video from over 10,000 vehicle platforms. In short, network-connected vehicles are not cameras, but AI systems need just several minutes to do this. AI restricted by line-of-sight limitations. Essentially, data on acci- cameras can also trigger alarms upon recognizing anyone dents, congestion/traffic jams and road blockages could be behaving suspiciously at big events, such as the 2016 G20 transmitted via 5G. Summit in Hangzhou. l Unleashing the power of data. The key finding from our work and interviews is that autonomous vehicles will generate data We are expecting smart security to be one of the top three growth on an unprecedented scale. Essentially, even with the contin- drivers of smart city spending, of which the majority will be in the uous evolution of technology, local processing power within public sector. vehicles is probably insufficient, meaning the use of cloud computing will become essential. V2X adoption l Back-up/contingency. We believe 5G will prove complementary to both V2V and V2I technologies. Indeed, we believe 5G Continuous penetration of autonomous vehicles requires signif- can be used as support in the event of slow networks (capacity icant connection to the network. V2X (vehicle-to-everything) is the blockages) and outages (power, cyber attacks). This would be new generation of communications technology that connects vehi- similar to the way that wireless networks and WiFi interact. cles to other vehicles, infrastructure, pedestrians, and networks. The

MORGAN STANLEY RESEARCH 43 M BLUEPAPER

Exhibit 56: Summary of the different connectivity platforms for AVs

Source: Morgan Stanley Research

Exhibit 57: Overview of connected car ecosystem

Source: Qualcomm, Morgan Stanley Research

44 M BLUEPAPER

Rollout of V2X in China is enabled by policy Exhibit 58: push and technology readiness. China has a Architecture of Vehicle-Infrastructure Cooperative Systems (VICS) good foundation in the development of V2X thanks to the scale of its automobile market, solid highway and telecom infrastructure, and comprehensive value chain. Several industry associations have started working on setting industry standards, including application definitions, technology requirements and security. The Chinese government has established multiple trial programs to explore the commercialization of V2X and business models together with all value chain players.

The automobile supply chain's focus has also been shifting to smart road/vehicle-infrastructure cooperative systems (VICS) in recent years. Along the V2X value chain, there are communication chipset/sensor module/ end-equipment makers, auto OEMs, operators, and service providers. Major suppliers that have already made related announce- ments include Hikvision, Navinfo, TUS, China Source: Morgan Stanley Research Transinfo, and Desay SV. Exhibit 59: We believe the commercialization of V2X will V2X trials in China follow these stages: Trial place Related parties Key applications Scale Shanghai Shanghai International Auto City Group Smart cars, V2X connections 18 roads and 3000 vehicles in 2018-19 Wuxi Ministry of Public Security, China Mobile, Huawei LTE-V2X information convergence 211 crossings and 5 highways Chongqing CAERI Smart cars already has 9.6km trial road 1. Establishment of connections. Beijing Beijing Innovation Center for Mobility Intelligent (BICMI) PCW (Pedestrian Collision Warning) already has 12km trial road Changchun FAW Security Warning, smart cars, smart transportation NA China has already started the trial of Hangzhou Local government Smart cars, smart transportation 34 LTE-V2X stations Wuhan Local government Autonoumous driving, smart transportation To build demonstration district in 5 years

this stage by upgrading road infra- Source: Morgan Stanley Research structure in areas of high automobile density. Users are starting to recog- Exhibit 60: nize and become interested in V2X Local governments are competing to build 5G-enabled pilot zones with basic applications pre-installed in cars. 2. Capability enhancement. Coverage expands along with penetration of commercial users. To support the enlarged user base and more new applications, network upgrades will become necessary, including multiple layers of computing capacity. 3. Application upgrades. ADAS will Source: Morgan Stanley Research. *PV=passenger vehicles, CV=commercial vehicles evolve to autonomous driving with the help of 5G-V2X technology. Cooperative traffic will also become available.

MORGAN STANLEY RESEARCH 45 M BLUEPAPER Shared mobility Exhibit 61: Major types of shared mobility Planning a smart city that delivers effective and equitable urban mobility solutions is one of the most pressing problems for cities throughout the world. Shared mobility is a business model innovation that improves the utilization rate of social resources and the efficiency of public transportation, such as online taxi hailing and shared cars/bikes. It also makes a meaningful contribution to reducing traffic congestion. Despite the different operating model, all shared-mobility solutions/apps have some common factors, such as: 1) reliance on mobile application to enable users to enter into rental/lease or usage contracts; 2) a strong social component for users to evaluate and share their experiences; and 3) customization of transportation services, to some extent. We believe shared mobility can enhance smart cities in terms of both environmental and social impact. l Environmental. We expect that car sharing will be a real opportunity for reducing car ownership levels, especially in Source: CIVITAS large urban areas. Bike sharing could also significantly reduce car use. We are also expecting car-sharing schemes to result in the replacement of old vehicles with more environmentally Remote healthcare: Leveraging IoMT technology, doctors can mon- friendly ones. itor patients' health remotely and analyze all of the data collected to l Social. By reducing the cost of transportation through sharing prescribe highly personalized treatments. Patients may even be able schemes, the prevalence of shared mobility services will to 3D print pills at home. Meanwhile, through data connectivity, doc- encourage more citizens to come out and travel, in our view. tors can also conduct remote surgeries. On 27 August 2019, doctors Moreover, the fast ramp-up of the geographical coverage of at a Beijing hospital successfully conducted a remote robotic surgery, shared-mobility services due to economies of scale could on a patient more than 136km away, using 5G wireless technology. quickly expand the boundaries of cities. Increasing hospital capacity: Through smart patient flow planning, China is a global leader in this kind of B2C business model inno- hospitals can improve the quality of services and the efficiency utili- vation. Though the capital-driven model, price competition and a zation of resources. Hospitals can connect beds or medical lack of regulation may distort the market in the short run, we believe machinery to improve efficiency. For example, GE partnered with a shared mobility will bring long-term value creation as it lifts the utili- hospital in New York to connect and track hospital beds using sen- zation rate of private cars, reduces matching times between taxis and sors. These sensors enabled hospital operators to tell when a bed passengers, and gives more choice to customers. was free and helped reduce emergency room wait times by as much as four hours, according to Business Insider. IoMT can also monitor Internet of Medical Things and remote healthcare the status of medical machines to reduce the probability of outages.

Smart wearable devices and IoMT: People can use wearable devices Smart ambulances: Smart ambulances can send a patient's data col- to access their health status or fitness regime without any profes- lected by sensors to the hospital's database while on the road, and sional help. Such devices can help check blood pressure, body tem- traffic signals could be adjusted to ensure the ambulance arrives at perature, heartbeat, cardiovascular problems, vision quality, and the hospital as fast as possible. At the same time, staff at the hospital chronic ailments. People can monitor their health condition in real can work on preparing treatments before a patient arrives. time, and smart devices can flag abnormal patterns.

46 M BLUEPAPER

Exhibit 62: Apple's health platforms and tools provide the foundation for a healthcare ecosystem

Source: Morgan Stanley Research

Online education and adaptive learning Many commercial education products are already yielding statistically positive results by improving study performance through adap- With online education, quality educational resources will not be tive learning tools. For instance, DreamBox Learning's online restricted by location. Experienced teachers and high-quality content individualized instruction program and Carnegie Learning's interac- can be made available to all students nationally, especially together tive math tutoring software achieve statistically positive effects in with the government’s efforts to unify the gaokao test paper and improving study performance. upgrade the syllabus. Students from remote suburban locations will have similar access to experienced teachers and learning resources as China's government is currently promoting Educational students in urban areas. Digitalization 2.0, which emphasizes developing smart campuses, improving students' and teachers' use of online resources, and inte- With adaptive learning, learning efficiency can be greatly enhanced. grating nationwide educational databases. Under Educational Teaching and learning progress can be digitalized through online edu- Digitalization 2.0, we believe the sharing of excellent online educa- cation, and big data analysis can be used to form personalized tional resources and the promotion of online learning will be critical learning plans for students and teaching suggestions for teachers. to narrowing the gap in teacher quality between schools. Longer term, with more adaptive learning of teaching behaviors, AI teachers could also become available.

MORGAN STANLEY RESEARCH 47 M BLUEPAPER

Exhibit 63: Score growth in US Southwest Allen County schools after usage of Dreambox

Smart home apps are making life more secure and convenient Source: Shutterstock

Smart homes

Current smart home applications are mostly single purpose, e.g., for controlling lights, air conditioners, entertainment systems, etc. With Source: Dreambox IoT technology becoming more mature, smart home applications could serve several purposes at the same time, with multiple systems Exhibit 64: connected through a home automation hub, allowing users to create Dreambox: Percentage of students who tested as proficient and distin- combined actions for particular situations. For example, a saved com- guished in statewide exams mand 'returning home from work' might include turning on the lights and air conditioner, preparing dinner, and suggesting entertainment or exercise options based on data collected through sensors at home (sleeping hours, dietary records, etc.).

A smart grocery system could automatically order food if it detects a shortage in the refrigerator through smart cameras. Orders could be delivered by drones, and the refrigerator would be refilled automatically.

Source: Dreambox Smart home security and surveillance systems can enable people to lock doors remotely through mobile apps and signal any abnormal situation. People can monitor their home security while on vacation. Security systems can also dial emergency calls in case of fire or inva- sion. Cameras installed on the door can record the presence of any suspicious persons.

48 M BLUEPAPER l Society governance: Security information sharing Case study: Smart city development across various departments to improve emergency management and urban security. in Shenzhen l Digitalization: Accelerating integration of internet, big According to the China Smart City Assessment Report data and IoT to promote new business models and released by the Center for Informatization Study (CIS), industrial digitalization. l Shenzhen ranked #1 among Chinese smart cities in 2018. Environment: Improving energy-saving in public Historically, Shenzhen has focused on informatization buildings, accelerating the establishment of smart development and has been a pilot city for several transportation and realizing dynamic and real-time national informatization projects, including the Smart monitoring of land utilization and waste disposal. Cities Project in 2012. Further development: In 2018, the 'Shenzhen Smart City Front-runner in smart city development: Construction Plan 2018-20' put forward new guidance, Shenzhen targeted smart city development even before with the target of being a world-leading smart city by the official launch of the Smart Cities Project. In 2011, the 2020. 'Smart Shenzhen Development Plan 2011-20' was l One integrated support network system: A released to focus on informatization development. The comprehensive communication network combining project aims to use new technologies (converged monitoring and computing capacity, providing networks, smart sensors and computing power) to fundamental IT support and data collection. improve urban management and quality of life and l Two centers: (1) The data center is in charge of data promote industry upgrades. management and sharing resources and (2) the smart In 2012, Shenzhen was selected as a pilot city in the first city operation center focuses on trans-department batch of the Smart Cities Project. cooperation and decision support services. In 2016, Shenzhen set up a special work team for smart l Four applications: (1) Smart public services include city construction, emphasizing support from local government, education, healthcare, and community enterprises, especially technology giants such as ZTE, services; (2) smart security includes public security, Tencent and Huawei. emergence management and safety production; (3) In 2016, the 'Shenzhen Smart City Construction Plan smart city governance focuses on improving 2016-20' was published with the following guidance: transportation, environmental protection and water utilities; and (4) smart industry targets smart industrial l Public services: Building an integrated internet service parks and factories. network covering healthcare, education and community services.

MORGAN STANLEY RESEARCH 49 M BLUEPAPER

Exhibit 65: Exhibit 66: Policy support in Shenzhen's smart city Shenzhen technology companies

construction Artificial Intelligence Big data and Cloud IoT

Smart Shenzhen Shenzhen set up a Shenzhen smart city development plan special work team for construction plan (2011-2020) smart city construction 2018-2020

2 2 2011 2012 2013 2014 2015 2016 2017 2018

Shenzhen was Shenzhen smart city selected in smart construction plan cities pilot project 2016-2020 Source: Shenzhen government data, Morgan Stanley Research Source: Morgan Stanley Research

Exhibit 67: Exhibit 68: Framework of Shenzhen's smart city in 2018- Smart city comparison

20 plan CityTarget Application/Programs Smart transportation Smart government Being a world-leading Shenzhen Smart services industry smart city Smart healthcare Smart community Smart transportation Smart government Building "Smart Hangzhou Smart safety Brains" for cities Smart healthcare Smart living Smart community Building a new smart Smart business district Shanghai city landmark Smart industry park Smart village Source: Shenzhen government data, Morgan Stanley Research Source: Morgan Stanley Research

Achievements: Smart healthcare: The healthcare department has developed a platform that enables citizens to have access Strong information infrastructure: Shenzhen has built a to various healthcare services, such as doctor robust broadband network with 100Mbps for individual appointments, hospital information, research, and users and 1000Mbps for enterprise users. Free WiFi is medical consultation, through mobile apps. The available in public areas, and IoT applications have been installment of WeChat payment for medical insurance in developed for utility meters. Shenzhen’s hospitals helps to save an average of 46 Smart government: Different departments are required minutes of waiting time. to co-build an integrated service platform to provide Smart transportation: Shenzhen's transportation services, with a goal of saving investment and improving department teamed with Huawei to build a efficiency through synergies. Citizens have access to the transportation system with a 'Smart Brain' to improve integrated platform through websites and mobile apps. In traffic safety and congestion. The system can recognize 2016, the resource-sharing platform integrated 385 traffic violations through algorithms and AI-based facial categories of information and resources of 29 units, recognition. Shenzhen has built a camera-and-sensor including over 3.8bn terms of data. 49 directly affiliated network to monitor license plates, accidents and traffic municipal departments and all district governments flows, collecting more than 700mn records per month. shared information via this platform. More than 90% of Coupled with big data analysis, the transportation government data was shared on the platform, and the department can reduce traffic jams by optimizing traffic utilization rate of shared resources was around 80% in flows. Also, the transportation system is linked to the 2017. integrated platform, which allows citizens to deal with mobile parking payment and online processing of traffic violations and fines.

50 M BLUEPAPER

Exhibit 69: Exhibit 70: Smart services via mobile apps Integrated services in mobile apps

Search for services

Navigation of Government various services to services like individuals and healthcare, enterprises security and transportation

Frequently used Healthcare government services services

Real-time updates on Doctor's appointments, bus transportation hospital locations, Automated External Def ibrillator (AED) etc.

Source: Morgan Stanley Research Source: Morgan Stanley Research

Exhibit 71: Exhibit 72: Smart transportation services Access to medical insurance via WeChat Mini-program

Transportation services

Online tickets; Real- Payment for medical time updates on bus insurance and details transportation of personal insurance information

Online processing of traffic violations and fines; Parking availability information

Traffic information Source: Morgan Stanley Research

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 51 M BLUEPAPER Initiative #3: Agricultural Modernization

Overview Moving rural workers into cities while preserving China's food security is critical to urbanization. Today, the key hurdles to China's agricultural modernization are uncertainty over land use and a fragmented and small-scale farming model. That said, we believe this can be resolved by the government's ongoing land reform efforts that started in 2014, as well as wider adoption of smart farming equipment such as drones, automated irrigation systems, and precision seeding equipment.

Key forecasts We expect China's agricultural labor productivity to more than double over the next decade, releasing more of the rural population who can then contribute to further urbanization.

Why is it important to enhance Exhibit 73: China's agricultural labor productivity remains low... agricultural productivity? Canada 93,110 Australia 85,075 United States 79,108 In our view, the key to unleashing more rural man- European Union 32,007 power for urbanization while preserving China's Japan 23,954 South Korea 19,113 food security is enhancing productivity. While Malaysia 18,112 China's agriculture productivity has increased at an Brazil 13,230 South Africa 12,025 8.1% CAGR over the past decade, from just 1.6% in Singapore 5,930 1969-78, on the back of reforms (see following box), Mexico 5,694 World Agricultural Labor Productivity in its level remains lower than that in most key econo- China 3,653 2017 (2010 US$) World 3,331 mies. Over the past few years, the country's trade def- Philippines 2,578 icit in agricultural products has been widening as a Vietnam 1,166 result of rising domestic food demand and a shrinking 0 20,000 40,000 60,000 80,000 100,000 area of arable land. With the UN estimating that Source: World Bank, Morgan Stanley Research China's population will continue to increase until Exhibit 74: 2030, this means a steady increase in food demand ...and its agricultural trade deficit is widening and continued downward pressure on the agricul- 60 USD Bn tural trade balance, suggesting the urgency of 40 boosting agricultural productivity. 20

-20

-40 Agriculture Imports -60 Agriculture Exports Agriculture Trade Balance -80

-100 1994 1998 2002 2006 2010 2014 2018 Source: CEIC, Morgan Stanley Research

52 M BLUEPAPER

China's agricultural policy reforms since 1978

China moved away from the collective agricultural system to the household registration system (HRS) in 1978-85. This provided more incentives for farmers to improve production, as HRS allowed farm households to either consume or sell any production beyond a fixed quota to be delivered to the country. To improve the demand-supply dynamics of agricultural products, policymakers also conducted market-oriented price reforms in 1985-93, rescinded SOEs’ monopolies on the trading of strategic products (e.g., cotton, soybeans) at the end of the 1990s, and further lowered tariffs on strategic agricultural products after China joined the World Trade Organization in 2001.

Exhibit 75: Timeline of China's agricultural policy reform

Source: Morgan Stanley Research

How can China's agricultural sector modernize?

In our view, uncertainty over land rights and the existing fragmented modern technologies, and this can also result in pollution and and small-scale farming model have been the key challenges in food safety issues. Exhibit 76 , for instance, shows that China’s recent decades, discouraging private investment, reducing the bene- pesticide application rate is among the highest in the world. fits of economies of scale, and slowing the adoption of new tech- Moreover, the inherent business risk from land rights may nology: dampen private investment in the agricultural sector. l Uncertainty over land use: In China, land is characterized by l Small-scale and scattered farms: China’s agricultural struc- collective ownership, and farm households can only use land ture is significantly smaller-scaled than global counterparts. on a leased basis. Despite the adoption of the household regis- This means it is difficult for farms to benefit from economies of tration system, uncertainty over land usage discourages scale because of a lack of communication and cooperation. farmers from preserving and protecting land resources by Meanwhile, it imposes challenges on new technology adoption, replacing poor quality farm chemicals with more expensive which is capital-intensive and largely unaffordable for small farms, and information gathering for ICT data analysis (there are 40,000 agriculture-related websites nationwide, which have fragmented and incompatible data).

MORGAN STANLEY RESEARCH 53 M BLUEPAPER Exhibit 76: Exhibit 77: Overuse of pesticides in China Relatively small farm sizes in China

China 13.06 Canada 331.8 South Korea 12.04 Japan 11.41 US 177.7 South America 4.85 EU 3.14 UK 79.8 US 2.63 World 2.57 Japan* 2.5 Mexico 1.87 Global Pesticide Usage in Average Farm Size in 2016 Canada 1.56 Agriculture in 2016, kg/ha South Korea* 1.5 (Hectare/Farm Households) Australia 1.1 South East Asia 0.89 China 0.7 Africa 0.31 0 2 4 6 8 10 12 14 0 50 100 150 200 250 300 350 Source: FAO, Morgan Stanley Research Source: USDA, Statistics Canada, China NBS, Statistics Korea, Japan Agriculture Ministry, UK Agriculture Department. Note: Data in 2015 for Japan and South Korea

In this context, we believe continued land reforms and the wider adoption of advanced agricultural technologies hold the key to boosting large-scale farming and modernizing the sector:

1. Continued land reforms

China has taken up a series of measures to facilitate large-scale farming. Since December 2014, China has started piloting reforms to separate farmland ownership, contract rights, and operating rights, which makes it possible for farms to collect scattered lands for large- scale planting. In August 2019, China's top legislature also adopted a revision to the land administration law, which gives farmers more property rights, enhances the protection of basic arable land, Most of China's farms don't yet have the scale to justify costly automated farm technologies Source: Shutterstock extends the duration of existing farmland use contracts by another 30 years upon expiry, and improves the transparency of rural land requisition. Beijing has also been promoting different forms of agri- tions, flying drones with cameras provide a bird’s eye view of cultural business entities in recent years, such as family farms and plant health and pest conditions, and in-field sensors monitor farm cooperatives, to improve the fragmented and small-scale soil conditions, sunlight levels, temperature, moisture, and air farming model. quality. Using these tools saves time compared with manual field checks. 2. Wider adoption of smart farming l Smart analysis: Based on real-time farming data, big data analytics can provide predictive insights and help make timely deci- Since 2014 policymakers have focused on promoting smart farming, sions. For instance, smart farming apps can suggest optimal which is based on advanced technologies, such as IoT, cloud com- times for planting, whether more irrigation, fertilization, or puting, big data analytics, and automation, to increase the quality and weeding is needed, which fields need pest control, and whether quantity of agricultural products. After precisely measuring varia- crops are ready to harvest. tions in fields, smart farming techniques help form the best strategies l Smart control: Automated equipment can conduct work via and use automation to conduct actual farming work. Specifically, remote control. For example, drones can be used for large-scale smart farming is concentrated on three areas: pesticide spraying and seeding from the air, precision seeding equipment can plant seeds at the correct depth and with the l Smart monitoring: Use of satellites, drones, cameras, and in- appropriate spacing to allow for optimal growth, and robots field sensors to enable real-time monitoring and geographic can water, fertilize, cultivate, harvest, and sort. data collection. For example, satellites provide weather predic-

54 M BLUEPAPER

In our view, the development of 5G and IoT, AI, and Exhibit 78: cloud computing will further integrate these three How does smart farming work? areas of smart farming, making full automation possible over the medium term. For instance, on a smart farm with an IoT-enabled monitoring system and automated equipment, when in-field sensors detect insufficient moisture levels in the soil, automated irrigation systems can immediately apply more water. Similarly, pest control and harvesting can be carried out without the need for human intervention. This would improve productivity and change the labor-intensive nature of China’s agricultural sector. Higher agricultural productivity to support further urbanization Source: Morgan Stanley Research

Still some room for the rural population to migrate to urban areas... The aging rural population and slower growth in migrant workers over the past decade has led to market concerns about the future potential for urbanization ( Exhibit 79 ). However, an NBS survey shows that there are still about 400mn rural people aged 5-50 (70% of the rural population) who will become or remain working age over the next decade ( Exhibit 80 ). This points to room for further urbanization, albeit at a slower pace than in the past.

...which needs to be enabled by higher agricultural productivity: International experience suggests that the share of agriculture in GDP tends to decline with higher GDP per capita as a consequence of industrializa- tion and developments in service sectors, and higher agricultural productivity is needed to meet rising domestic food demand ( Exhibit 81 ). In the US, Japan and Korea, agricultural labor productivity increased exponentially when the share of agriculture in GDP fell to 3-4% ( Exhibit 82 ). Our econometrics model suggests that when China’s per capita income increases to US$17,800, a level we forecast it will reach by 2030 (vs. US$9,450 in 2018), the share of agriculture in GDP would fall to 2% from 7% today. Assuming that China follows a path similar to that of developed countries, through successful implementation of rural reforms, inviting more private investment and FDI, and the adoption of smart farming, we believe China’s agricultural productivity could more than double over the next decade, releasing more rural workers for further urbanization.

MORGAN STANLEY RESEARCH 55 M BLUEPAPER

Exhibit 79: Exhibit 80: Despite an aging rural population and slower growth of migrant ...there is still a rural population of 400mn aged 5-50 workers... Rural Population at the age of 5-50 73% Person mn-RS 500 51% Rural Age Dependency Ratio-LS 6% % of Rural Population-LS 450 Growth of Migrant Workers-RS 72% 400 49% 5% 71% 350 70% 300 47% 4% 69% 250 200 45% 3% 68% 150 67% 43% 2% 100 66% 50 41% 1% 65% 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 39% 0% Source: CEIC, NBS, Morgan Stanley Research 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: CEIC, NBS, Morgan Stanley Research

Exhibit 81: Exhibit 82: Share of agriculture in GDP tends to decline with higher GDP per capita China's agricultural labor productivity would more than double by 2030 20 assuming it follows the path of the US, Japan and Korea 18 50,000 US (2000) Korea US 16 45,000 Japan China y = -4.9ln(x) + 50.4 14 40,000 R² = 0.7 35,000 12 China Korea 30,000 (2017) 10 Japan 25,000 8 (2014) China Agriculture Agriculture in Share GDP 20,000 (2030E) 6 Korea 15,000 (1992) 4 China 10,000 (2018)

2 Agriculture LaborProductivity, PPP 5,000 Japan (1980) 0 0 US (1950) 0 10000 20000 30000 40000 0% 5% 10% 15% GDP Per Capita, USD (International Dollar PerPerson, 2010 Price) Data as of 2017 Agriculture Share in GDP Source: World Bank, IMF, Morgan Stanley Research Source: World Bank, IMF, Morgan Stanley Research estimates

56 M BLUEPAPER

Micro Implications

MORGAN STANLEY RESEARCH 57 M BLUEPAPER Investment Theme #1: From a Consumer to an Industrial Internet

Exhibit 83: Investment Theme #1: Summary

Key Beneficiary Key 2030 Forecasts Top Stocks

5G capex: US$400bn in 2019-30 Telecoms 5G infrastructure companies (2x 4G capex) • China Tower (0788.HK)

• Alibaba (BABA.N)

• GigaDevice Semiconductor Technology leaders that are Beijing (603986.SS) Software and IT services spending: Internet expanding from consumer to US$200bn (5x 2018) industrial applications • HIKVision Digital Technology (002415.SZ)

industrial internet industrial • Yonyou Network Technology (600588.SS) Top players geared to IoT and 5G; Total value of connected devices: Tech Hardware and software vendors focusing on digital • VenusTech (002439.SZ)

Theme 1: From a consumer to an a consumer 1: From Theme US$684bn transformation or with smart city Software (vs. US$301bn in 2018) exposure

Source: Morgan Stanley Research

58 M BLUEPAPER 1a. Telecoms

Overview: China's extensive fiber network and pioneering 5G rollout are the major enablers of smart cities. Globally, telecoms operators are struggling to identify business cases to monetize 5G investments, but we believe smart cities will emerge as an early B2G application.

Key forecasts: We project 5G capex of US$400bn in 2019-30, which is double the level of 4G capex. We forecast IoT connections (connected with carrier networks) will increase 10x, from 800mn in 2018 to 8bn in 2030.

Investment China Tower is our preferred 5G play, while China Communication Services should be a key beneficiary of smart implications: city development.

China's telecoms infrastructure ranks among the best in the world, Fiber: China is one of the most fiberized countries globally. As of June and it includes extensive fiber coverage and a pioneering 5G rollout, 2019, according to MIIT, total optical fiber length reached 45mn km. paving the way for smart city development. We project total capex Fiber-to-the-home/office (FTTH/O) subscribers reached 400mn, of US$400 bn in 5G (including transmission) in 2019-30. Although representing more than 90% of broadband subscribers in China. 5G use cases remain immature today, Chinese telcos tend to focus on Many smart home applications, such as connected home appliances, to-business (2B) and to-government (2G) industrial applications, and surveillance and entertainment, require extensive FTTH penetration. we believe smart cities could emerge as one of the early 5G monetiza- In China, almost all of the mobile base transceiver stations (BTS) are tion opportunities for operators, given policy support. Among the connected with fiber for transmission, which is essential to providing three Chinese telcos, we believe China Telecom (CT) and China 4G/5G services. Unicom (CU) are better positioned than China Mobile (CM), given higher exposure to the government/enterprise segment and net- l 'Broadband China' policy: In 2013, the State Council work sharing. For telecoms infrastructure, we believe China Tower announced its 'Broadband China' plan, setting 2020 targets for (CTC), China Communications Services (CCS) and data centers are the number of broadband subscribers, network speed, capacity, the key beneficiaries. and penetration. Most of the targets were aggressive compared to the broadband network at the time. However, as of June 2019, most of the targets had been exceeded, making China one of the most fiberized countries globally. l 'Speed Upgrade Tariff Reduction' initiatives: Since 2015, the State Council has been setting 'Speed Upgrade Tariff Reduction' targets, which lower broadband tariffs substantially for both household and enterprise subscribers, facilitating the penetration of the fiber network.

MORGAN STANLEY RESEARCH 59 M BLUEPAPER

Exhibit 84: 'Broadband China' targets

Broadband China Targets Index Unit 2013 2015 2020 Jun-19 1. Broadband users Broadband subs Mn 210 270 400 435 FTTH/O subs Mn 30 70 NA 398 Urban subs Mn 160 200 NA 306 Rural susb Mn 50 70 NA 129 3G/4G Subs Mn 330 450 1200 1586 2. Broadband penetration Broadband penetration % 40 50 70 94 Urban subs % 55 65 NA NA Rural susb % 20 30 NA NA 3G/4G subs penetration % 25 32.5 85 114 3. Broadband capability Urban Area Mbps 2080% subs 20 50 100 100some 1000some Developed cities Mbps NA 1000 (some cities) cities subs Rural Areas Mbps 485% subs 4 12 NA Speed for enterprise Mbps >100 >1000 NA FTTH subs Mn 130 200 300 344 3G/LTE BTS '000 950 1200 NA 4450 Broadband penetration in % 90 95 >98 >98 administrative villages 4. Broadband information application Internet users Mn 700 850 1100 854 Rural users Mn 180 200 NA 318 E-commerce transaction Bn 1000 1800 NA 3660 (by 2018) volume

Source: State Council, MIIT, Morgan Stanley Research

Exhibit 85: 'Speed Upgrade Tariff Reduction' targets

Source: State Council, Morgan Stanley Research

5G: China has become the world's front-runner for 5G network buil- Low Latency Communications (URLLC) and Massive Machine dout, a key building block of smart city development. Type Communications (mMTC), which are mainly B2B use cases, facilitating smart city development. l Timetable: 5G spectrum was allocated in December 2018, with l Use cases: Encouraged by the government, telcos have been 3.5GHz for CU and CT (100MHz each) and 2.6/4.9GHz for CM conducting trials for various use cases, especially for smart (160MHz and 100MHz, respectively). The sufficient sub-6GHz cities, such as smart transportation, manufacturing automation, spectrum is well positioned for 5G development. 5G licenses remote education and healthcare, autonomous driving, and were then released in June 2019. Telcos could launch commer- smart homes. cial 5G services as early as late-2019. l Network slicing and edge computing: Smart cities may need l Capex: We expect Chinese telcos to spend total 5G capex network slicing owing to cybersecurity requirements, which (including on wireless networks, core networks, transmission, enables telcos to exert higher pricing power. This technology and towers) of Rmb2.8trn (US$400bn) in 2019-30. enables telcos to sell 'slices' of the network, at different param- l SA vs. NSA: Unlike most other countries, which started 5G net- eters, customized for specific end-applications. Meanwhile, work deployment under a non-standalone (NSA) model, China emerging technologies, such as edge computing, enable smart insists on developing its 5G networks using a standalone (SA) city applications to run smoothly on a city-wide basis. model, preparing telcos for 5G applications in Ultra Reliable

60 M BLUEPAPER Infrastructure sharing: The Chinese government has been sup- CT and CU are better positioned than CM: Telcos have been strug- portive of infrastructure sharing, which we believe will improve the gling to identify use cases to monetize 5G capex investment. efficiency of smart city development. Consumer applications are more mature, but monetization is difficult given a lack of pricing power; enterprise applications potentially l Tower sharing: Encouraged by the government, in 2014-15 offer significant upside, but applications are very immature. We China Tower (CTC) was established to take over new tower believe initial use cases will be government-driven applications construction from the three telcos and start a tower-sharing focused on smart cities and smart agriculture. Currently, 60-80% of scheme: CTC meets telcos' new tower demands by using other Chinese telcos' revenues come from the consumer market, and we telcos' existing towers. CTC then acquired all existing tower believe the telcos are well positioned to benefit from incremental assets from the three telcos and improved network efficiency revenue from enterprise and government, given supportive policies, through co-locating tower demands from telcos by offering co- which, in our view, is not reflected in the share prices. location discounts. l DAS sharing: The government encourages CTC to take leader- CU's and CT's plan to build a nationwide 5G network is a significant ship in the coordination of indoor distributed antenna system positive for both companies, given the substantial capex savings. CU (DAS) deployment. CTC negotiates with local governments to and CT indicated that the combined 5G capex will be similar to that acquire social resources (e.g., light poles) and consolidates the of CM, even with network sharing, which means CM's network advan- resources to meet telcos' demands, improving network rollout tage is likely to diminish during the 5G era. efficiency. l 5G network sharing: In September 2019, CU and CT together announced they will build a full-scale nationwide 5G network, which we believe is positive for smart city development given the efficiency of co-building networks.

Exhibit 86: CU and CT's 5G network co-build plan CU-CT network co-build co-share

Ratio of construction Ratio of construction Region Province/City districts or sole builder Region Province/City districts or sole builder 15 Cities 25 Provinces Beijing Hebei Tianjin Henan 5 Northen Cities Zhengzhou CU:CT = 6:4 Heilongjiang Qingdao Jilin 8 Northern Provinces CU Shijiazhuang Liaoning Shanghai Inner Mongolia Chongqing Shandong Guangzhou Shanxi Shenzhen Anhui Hangzhou Fujian 10 Southern Cities CU:CT = 4:6 Nanjing Gansu Suzhou Guangxi Changsha Guizhou Wuhan Hainan Chengdu Hubei Guangdong and Zhejiang Hunan CU: 9 prefecture cities Guangdong Province / CT: 10 prefecture cities 17 Southern Provinces Jiangsu CT CU: 5 prefecture cities Zhejiang Province / CT: 5 prefecture cities Jiangxi Ningxia Qinghai Shaanxi Sichuan Xizang Xinjiang Yunnan

Source: Company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 61 M BLUEPAPER Exhibit 87: Exhibit 88: China telcos: Enterprise revenue contribution is still relatively small China telcos: CU and CT together spent less capex than CM CM 2018 Service CU 2017 Fixed-line during the 4G rollout (2014-18) Revenue Structure Broadband Access Revenue Rmb bn Emerging Business CM CT CU 450 419 9% Internet Dedicated Corporate Market Line Access 400 348 11% Revenue 350 27% 300 Thousands 240 Household Market 250 187 7% 200 144 150 48 Personal Mobile 100 Market Others 160 73% 73% 50 96 - Mobile Transmission Mobile Transmission Source: Company data, Morgan Stanley Research Source: Company data, Morgan Stanley Research

China Tower: CTC is a key beneficiary of 5G investment and smart Exhibit 89: city applications. Its tower business should benefit from the com- China Tower: EBITDA and net profit EBITDA Net Profit mencement of the 5G capex cycle in 2H19, which requires denser (Rmb bn) China Tower: EBITDA vs. Net Profit (Rmb bn) 80 25 macro cells supplemented by small cells. Meanwhile, its DAS busi- Net Profit EBITDA 69.6 70 80% 2018-20 CAGR: Net Profit 62.1 2020-23 CAGR: 38% 20 ness will benefit from the indoor coverage needs of 5G applications. 60 55.1 49.1 Furthermore, we believe CTC's trans-sector site application and 50 44.7 41.8 15 40 information (TSSAI) business is well positioned to benefit from 22.6 30 10 smart city applications, e.g., surveillance, weather monitoring, traffic 17.4 20 12.6 data collection, etc., leveraging its tower assets. 8.6 5 10 5.5 2.7 0 0 2018 2019E 2020E 2021E 2022E 2023E Data centers: In our view, data center operators are well positioned Source: Company data, Morgan Stanley Research estimates to benefit from secular demand growth in computing and storage capability by providing infrastructure to public cloud vendors or gov- Exhibit 90: ernment and enterprises for their private IT deployment. The con- CCS: Smart city development struction of smart cities would further increase overall internet traffic and data volumes. In addition, the government's increasing focus on environmental protection, particularly in key city clusters, raises the hurdle for new entrants. We believe leading incumbents like GDS, Sinnet and 21Vianet will enjoy rising value from their existing capacity and a competitive advantage in future expansion.

China Communication Services: CCS is a major beneficiary of smart city construction. It has a deep understanding and rich experience in smart city construction. It has market-leading capabilities in top-level Source: Company data planning, turnkey ICT solutions, systems integration, operation and Exhibit 91: maintenance. In addition, close relations with municipal govern- CCS: Non-telecom revenue growth ments and a neutral platform for all partners make it even more com- 140

120 11% 5-year petitive in smart city-related tendering. The geographical expansion CAGR of smart cities and deepening of smart applications are key drivers of 100

CCS's revenue and profit growth, supported by the government's 80

high investment priority and fiscal budget. bn Rmb 60

- 2016 2017 2018 2019E 2020E 2021E 2022E 2023E

CCS Non-telecom revenue (primarily smart city and related)

Source: Company data, Morgan Stanley Research estimates

62 M BLUEPAPER 1b. Internet

Overview: China's internet giants will likely be key enablers of Urbanization 2.0, as they are technology leaders and are expanding from consumer to industrial applications (including cloud computing and smart solutions), which enable the development of smart cities and digitize enterprises in each vertical market.

Key forecasts: We project software and IT services spending will grow at a 13% CAGR, to US$200bn in 2018-30, with global spending share reaching 8% (vs. 3% in 2018).

We project online retail GMV to reach about Rmb30trn in 2030, rising at a 11% CAGR from 2018, with over 1bn online shoppers (vs. 610mn in 2018), thanks to urbanization driving up internet penetration and improved logistics helping e-commerce penetration in categories such as FMCG and fresh groceries, on top of an expanding internet population.

We project food delivery GTV will reach Rmb2.2trn in 2030, rising at a 13.5% CAGR from 2018, with average daily orders ramping up to 126mn (vs. 29mn in 2018), as we note China has more than 150 cities with a population of over 1mn (vs. only around 10 cities of such size in the US), where urban appetites should continue to fuel industry growth.

Investment Alibaba and Tencent should be key beneficiaries in view of cloud computing demand. Baidu is facing a bigger chal- implications: lenge in its core business due to the rise of Bytedance, so the potential benefits from the industrial internet could be partially offset by weakness in the core business. Urbanization should benefit physical goods and service e-commerce leaders in lower-tier cities, favoring Alibaba and Meituan given their market-leading positions and strong ecosystems.

In an era of slowing internet use growth by consumers, China's major internet companies are ramping up efforts to develop an industrial internet with respect to infrastructure investment (cloud), capabilities (AI, big data, industry applications), and organizational restructuring. Their digitalization initiatives aim to cover a wide range of sectors such as retailing, manufacturing, transportation, and government/smart cities by working with various business partners to form ecosystems. We believe internet companies are enabling various stakeholders (enterprises and governments) to enhance their capabilities to support further urbanization and create better household living in China.

MORGAN STANLEY RESEARCH 63 M BLUEPAPER

Exhibit 92: Alibaba’s industrial internet footprint

Source: Company data, Morgan Stanley Research

Exhibit 93: Tencent’s industrial internet footprint

Source: IDC, McKinsey Global Institute, Morgan Stanley Research

64 M BLUEPAPER Industrial internet carries ample potential...: China's government However, as it did with the consumer internet, we believe China is has announced several initiatives to upgrade the country's manufac- likely to leapfrog other countries in the industrial internet. Digital turing sector to monetize the sizeable amount of manufacturing data transformation in the US evolved in three major steps: IT > Cloud > available. Both the public and private sectors are focusing on technol- AI & big data, whereas in China these three development steps for ogies such as AI and automation to transform China into an innova- enterprises could happen at the same time. We expect Chinese tive, high-tech powerhouse. We think this is further enhanced by the internet companies to play a meaningful role in certain fields of following points: 1) the rising cost of human capital – wages have enterprise IT spending, including IT services (cloud infrastructure ser- grown by more than 10% annually over the past eight years; 2) vices, consulting, implementation, managed services and business maturing cloud service development – we expect the public cloud process outsourcing) and software (data analytics and business intel- adoption rate to move from 4% in 2016 to 16% in 2020; 3) the rapid ligence, supply chain, ERP, CRM and vertical industry-specific applica- adoption of mobile internet (60% of the population); and 4) the gov- tions). ernment's strategic focus should fuel the development of the 2B service market in China. China's software and IT services market has significant growth potential given that the country is underspending on software relative to Despite significant efforts by the government and enterprises in the the US ( Exhibit 95 ). According to Gartner, China's enterprise IT past couple of years, we believe there are still ample opportunities spending by segment (ranked from highest to lowest) is: telecom ser- for further digitalization in both the 2B and 2G segments, in that vices, devices, datacenter systems, IT services, internal services, and Chinese enterprises have relatively narrower adoption of digitaliza- software. In contrast, the ranking for US enterprises is: IT services, tion as compared with US and global enterprises. software, internal services, telecom services, data center systems, and devices. Therefore, we forecast China's software and IT services spending to grow to US$200bn in 2030.

Exhibit 94: Exhibit 95: China’s enterprise IT spending was only 18% that of US enterprises in 75% of China’s enterprise IT spending relates to hardware, with only 2018 25% related to software and IT services (US$bn) China Enterprise IT Spending vs. Internet Aggregated Revenue Software + IT Services as % of Enterprise IT Spending Others Data Center Systems Software + IT Services Internet 70% 64% 65% 350 61% 63% 322 59% 60% 60% 58% 300 58% 250 50% 55% 56% 52% 54% 250 50% 51% 210 197 40% 31% 200 169 178 177 29% 30% 25% 27% 22% 23% 150 21% 120 20% 100 10%

50 0% 2016 2017 2018E 2019E 2020E 2021E 2022E 0 IT Internet IT Internet IT Internet IT Internet World Wide US China 2016 2017 2018E 2019E Source: Gartner (E) estimates, Morgan Stanley Research Source: Gartner, Morgan Stanley Research estimates

Exhibit 96: Exhibit 97: China's software and IT services spending to grow at a 13% CAGR …accounting for almost 8% of global spending (up from 3% in 2018) through 2030... 10% (US$bn) Software IT Services 9% 7.6% 8% 7.1% 250 6.5% 7% 6.0% 5.6% 200 6% 5.1% 200 4.7% 180 4.2% 4.4% 5% 3.9% 161 3.6% 4% 3.3% 143 3.0% 150 127 3% 113 99 2% 100 87 76 1% 66 57 0% 45 50 50 2018 2019E 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E Source: Gartner, Morgan Stanley Research estimates 0 2018 2019E 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E Source: Gartner, Morgan Stanley Research estimates MORGAN STANLEY RESEARCH 65 M BLUEPAPER ...with monetization to come from the public cloud and other & Smart Industries Group to offer enterprise digitalization solutions infrastructure: We believe the benefits of digitalization will take across retail, healthcare, education, transportation, and municipal time to be realized and that efficiency improvements may initially be services. It aims to expand from IaaS to PaaS and SaaS over time, difficult to quantify. At the same time, smart cities rely on data while leveraging AI, location-based services, and payment among other the cloud will be a crucial platform and supplies resources/capabili- core technologies. Baidu is cultivating its new AI businesses (i.e., ties for data accumulation and processing. We therefore think infra- DuerOS, Apollo and AI Cloud) to tap into the enterprise solutions structure such as cloud computing/storage/platforms is likely to be market and smart city projects. Additionally, Baidu has said it wants developed first. to position its AI cloud beyond delivering infrastructure, and it targets to deliver data insights and solutions. We believe that one of the early successes of China's industrial internet development will be the public cloud market as enterprises Internet companies are facilitating the smart city concept: Smart shift their storage/computing/application solutions. The public city initiatives are complex projects, involving many interrelated cloud market in China was US$7bn in 2018 sales, according to IDC, domains such as transportation, public services, healthcare, security, making up merely 6% of the global market. This is well below the infrastructure (5G and cloud) and other capabilities (AI and block- 10-15% global value share for China's other tech products – PCs, chain). Chinese internet companies are serving as key enablers and smartphones and servers – implying low public cloud adoption (11%, pathfinders for such initiatives and have demonstrated initial prog- on our estimates). Nonetheless, we believe China is set for wider ress. public cloud adoption given government support, improving product quality, lower costs, and rising demand for new technology (AI, big Take, for example, Alibaba's City Brain project in Hangzhou, Zhejiang data, IoT). We project that China's public cloud market will reach province. In late 2016, Alibaba announced it was cooperating with the US$18.8bn by 2020, with public cloud penetration rising to 22% Hangzhou government to develop ET City Brain, an AI-driven project from 11% in 2018. helping the government better manage urban areas. In its 2019 Apsara (cloud computing) Conference, it illustrated City Brain's Additionally, information security concerns and the government's smart transportation system, which enables traffic light adjustment efforts to have key products and components 'made in China', should and allows drivers to adjust travel routes based on real-time traffic be an important tailwind for domestic players like AliCloud. We fur- predictions, alleviates road congestion in Hangzhou, which was once ther analyze the competitive dynamics for BAT's cloud businesses: one of the top five most congested cities in China and now has Alibaba: Among China's public cloud companies, we think Alibaba is improved to the national average level. It also helps with the integra- a clear leader given: 1) its sheer scale, from leveraging third-party tion of data platforms and systems across multiple government datacenters to building in-house, 2) its demonstrated ability to agencies, which speeds up public services. In the past, citizens in expand, as shown by the Double 11 event, 3) data security concerns Hangzhou had to experience a time-consuming process with a should benefit domestic competitors, and 4) it is expanding its number of visits to various government agencies in different loca- product offerings from IaaS to SaaS. Tencent has established a Cloud tions get the required documents and procedures done for just one

Exhibit 98: Exhibit 99: China’s public cloud market made up only 4% of the global public cloud We project that China's public cloud market will reach US$18.8bn in market in 2018 2020, with penetration rising to 22% Public cloud revenue in China (LHS) (US$ bn) Public cloud adoption rate (RHS) YoY (RHS) Global public cloud ~US$183bn 20 18.8 100% 78% 80% 15 64% 66% 56% 57% 60% 12.0 US public cloud ~US$110bn 10 7.3 40% 22% 5 4.1 16% 2.5 8% 11% 20% 5% China’s puli loud ~US$7n 0 0% 2016 2017 2018 2019E 2020E Source: IDC, Morgan Stanley Research Source: IDC, Morgan Stanley Research estimates

66 M BLUEPAPER application. Now, citizens can get about 90% of public services ful- expansion into segments such as fresh groceries as major players filled with at most one physical visit to a local government agency. improve their e-commerce infrastructure. We forecast e-commerce GMV to reach Rmb31trn in 2030, implying a 11% CAGR from 2018. E-commerce penetration to be driven by urbanization: We estimate that the e-commerce penetration rate (e-shoppers/total popu- Urban appetites to fuel food delivery industry growth: We lation) was 34% in low-tier cities and rural areas in 2018 vs. 71% in tier believe e-commerce platforms for services (i.e., food delivery, in- 1 and 2 cities, given low internet penetration and less-developed store services) should continue to benefit from the urbanization logistics infrastructure. Low penetration in less-developed regions theme. We estimate food delivery GTV to reach Rmb2.2trn in 2030, resulted in a mere 44% overall e-commerce user penetration in rising at a 13.5% CAGR from 2018, with average daily orders ramping China, vs. 60% in the US. We believe continuous urbanization along up to 126mn (vs. 29mn in 2018), followed by higher purchase fre- with improved infrastructure will boost the rate to 77% by 2030, or quency on a per-user basis. an online shopping population of over one billion. E-commerce's wallet share is also expected to increase, driven by product category

Exhibit 100: Exhibit 101: Online shopping population to reach 1bn by 2030, driven by urbaniza- E-commerce GMV to Rmb31trn tion and improved infrastructure (Rmb bn) 35,000 (mn) 31,151 1,200 1,095 30,000

1,000 25,000

20,000 800 610 15,000 600 8,865 10,000 400 5,000

200 0 2018 2030E 0 Source: iResearch, Morgan Stanley Research estimates 2018 2030E Source: CNNIC, Morgan Stanley Research estimates

Exhibit 102: Exhibit 103: Food delivery average daily transactions to reach 126mn by 2030, Food delivery GTV to reach Rmb2.2trn by 2030 driven by urbanization and higher purchase frequency 2,500.0 China Food Delivery GTV (Rmb bn) Average daily transactions (mn) 2,187.6 160.0 2,000.0 140.0 125.5

120.0 1,500.0 100.0

80.0 1,000.0

60.0 479.0 500.0 40.0 29.2

20.0 0.0 2018 2030E 0.0 Source: iResearch, Morgan Stanley Research estimates 2018 2030E Source: Company data, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 67 M BLUEPAPER Stock implications Tencent l Reorganization to expand from consumer to industrial We believe the industrial internet presents ample growth potential internet: In September 2018, Tencent announced an important for early movers in the 2B and 2G segments, including Alibaba, reorganization to enable it to embrace the consumer and indus- Tencent and Baidu, and we expect them to be the major beneficiaries trial internet. It set up a dedicated Cloud and Smart Industries when industrial internet investments begin to bear fruit. Each has Group, as it now aims to pursue the growth potential of the successful cash cow consumer internet businesses to support early industrial internet. It also plans to set up a technology com- investment in the industrial internet. However, these investments mittee, pulling in internal resources and enhancing fundamental will affect their profitability, especially as it will likely take a long time research. l for profits to materialize. China's second-largest public cloud vendor: Tencent has been strategically focused on its public cloud business, with In terms of their positioning in the data era, we believe Baidu has a strong growth momentum from a low base. It has grown to better technological foundation, given its core business in search become China's second-largest public cloud vendor, with engines. It has specified two strategic focuses in AI: 1) autonomous around 12% IaaS revenue market share in 2H18, per data from cars and 2) DuerOS, which is an operating system for IoT products. IDC. l Nonetheless, Baidu is likely to face structural headwinds in the online Tapping into 2B business by leveraging success in 2C: ad business because of the rise of Bytedance, so its core business Tencent has emphasized that it believes the industrial internet growth could be much slower than either that of Alibaba or Tencent. will provide the next wave of internet opportunities. It will On the other hand, in terms of user scenarios, we believe Alibaba and focus on offering integrated solutions to digitalize enterprises Tencent are better positioned, given their richer user and merchant across retail, healthcare, education, transportation, manufac- databases. In view of this, we expect BAT to excel in different areas turing, and municipal services. To do this, Tencent plans to in the data era. leverage its core AI, big data, cloud, and mini program technologies to bring offline enterprises onto the internet. Alibaba Baidu l Data technology as the key edge: We believe Alibaba's l closed-loop ecosystem provides good quality data for analytics. Top-line growth slowdown amid structural challenge in We also like Alibaba's new retail strategy as a 2B initiative, as it online ad and macro risks: Baidu may not be immune to aims to improve the efficiency of online and offline retailers. macro risks, and the top-line growth boosted by the Baidu App l China's largest public cloud vendor: Alibaba's public cloud newsfeed and iQIYI business lines are likely to gradually slow in services commanded 43% market share by IaaS revenue in 2019. In addition, the rise of competitor Bytedance has created China in 2018, according to IDC, which we think lays a solid structural challenges for Baidu and other apps. We thus expect foundation to grow services for industrial applications. In addi- Baidu to experience slower revenue growth than its China tion, it is expanding services from IaaS to PaaS/SaaS by pro- internet peers until new mobile initiatives (i.e., short videos – viding digitization solutions for enterprises, which could Haokan video app) start to make meaningful progress, and the provide future growth. commercialization of new AI businesses (i.e., DuerOS, apollo, AI l Increased investments in new technology: Alibaba set up Cloud) takes place. l DAMO Academy to invest in future technologies, and it also set Well positioned, but patience needed: We believe Baidu is up a semi company to develop proprietary AI chipsets in 2018, well positioned for the enterprise solutions market and smart demonstrating its ambition to lead in this new area, which has city projects, backed by its 2B business, data edge, technology developed and released CPU IP (Xuantie 910), an SoC chip plat- roots and dedicated investments. Baidu has also seen progress form (Wujian), and an NPU (Hanguang 800) in 2019. in certain fields (i.e., autonomous driving, IoT ecosystem) with commercialization on the way, which we view as a long-term positive. We believe Baidu will be able to capture 2B/2G opportunities post the ecosystem buildup.

68 M BLUEPAPER Meituan Online recruitment l Urban appetites to fuel food delivery industry growth: Meituan is China’s leading e-commerce platform for services Although we forecast the total employed population in China to capitalizing on the booming food delivery market. We believe remain stable, at around 770-780mn through 2030, urbanization Meituan should continue to ride the urbanization wave in will lift the total number of urban employees from 426mn in 2017 to China. Take its core business food delivery, for example. The 585mn in 2030, accounting for 54% of the employed population, per China market has higher growth potential than overseas mar- our analysis. kets thanks to China's higher population density and consumption upgrade trend. In addition, higher order volumes should According to the China Internet Network Information Center, China's drive better efficiencies, leading to higher profitability over the mobile internet users reached 817mn in 2018 thanks to the improving long term. penetration of low-cost smartphones. We expect online recruitment l Traffic redirection to drive future monetization: Meituan to keep increasing on the back of mobile penetration. According to has effectively leveraged its success in high-frequency food ser- iResearch, the number of online job seekers will increase 8%, to vices to expand into low-frequency services (i.e., hotels, travel) 193mn in 2019, despite the weak macro economy, and we expect on its one-stop platform. Moreover, thanks to its industry lead- online job seekers to further increase to 316mn in 2030, representing ership, Meituan’s logistics infrastructure and merchant solu- 41% of the total employed population. tions are superior to peers, in our view. We believe Meituan will continue to empower merchants in a fragmented service supply chain.

Exhibit 104: Exhibit 105: Number of urban employees Number of online job seekers (mn) Number of urban employees % of total employment (mn) Number of online job seekers % of total employment 700 80% 350 45%

600 75% 300 40% 70% 500 250 35% 65% 400 200 30% 60% 300 150 25% 55% 200 100 20% 50% 100 45% 50 15% 0 40% 0 10% 2019E 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E 2013 2014 2015 2016 2017 2018

Source: NBS, Morgan Stanley Research estimates Source: iResearch, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 69 M BLUEPAPER In addition, we expect strong job demand in the service industry to Further, according to the China Federation of Logistics & Purchasing, persist. Despite near-term macro headwinds, demand for workers the YoY growth of the E-commerce Logistics Index rose steadily to from O2O and logistics remains resilient. According to iResearch, 8% in July 2019 following a bumpy period starting in November 2018. online food delivery accounted for 16.3% of the total O2O market in The recovery indicates a positive trend in the development of e-com- China in 2018, showcasing the rapid growth of the past five years. merce amid headwinds in the economy; as a result, demand for blue- Analysys projects that the total market size of online food delivery collar workers from the e-commerce logistics industry should will expand to Rmb934bn in 2021, with a three-year sales CAGR of remain resilient, in our view. 28%, underpinning continuing demand for workers.

Recently, Alibaba's local service arm launched an open platform to digitalize brick-and-mortar retailers, and it already covers more than 10,000 large-sized supermarkets and nearly 200,000 chain stores in 676 cities. The expansion of delivery service from catering to other segments is also driving up demand for blue-collar workers.

Exhibit 106: Exhibit 107: Online food delivery market is expected to grow Recovery in E-commerce Logistics Index in 1H19 (mn) Number of online job seekers YoY (RHS) 200 20% 193 21% 179 180 161 19% 160 144 17% 140 129 115 15% 120 104 94 100 87 12% 12% 12% 13% 78 11% 10% 11% 80 65 11% 60 12% 8% 8% 9% 40 20 7% 0 5% 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019E Source: Analysys

Source: China Federation of Logistics & Purchasing, Morgan Stanley Research

70 M BLUEPAPER 1c. Tech Hardware and Software

Overview: Asia's tech companies are key enablers of Urbanization 2.0, as smart cities require security solutions, IoT networks comprising sensors, MCUs, communication ICs and power ICs, and extensive software for data management and analysis.

Key forecasts: The total value of connected devices should more than double from US$301bn in 2018 to US$684bn in 2030, we estimate, with the major driver being public sector infrastructure investment. The household segment will eventually outgrow the public sector, given low penetration currently. Meanwhile, the total market size of software and IT services could grow fivefold, to US$200bn by 2030, per our projections.

Investment Tech companies geared to IoT and 5G should benefit from Urbanization 2.0, including TSMC (leading-edge semi implications: foundry), MediaTek (5G and IoT chip supplier), Win Semi (5G PA foundry), Realtek ( connectivity chip), Will Semi (sensors), ASM Pacific (volume play for semis), Macronix (NOR), GigaDevice (NOR+MCU); HIKVision (surveillance), ZTE and Accelink (telecom equipment); Foxconn Industrial Internet (hardware + software solutions).

We believe software vendors will become major enablers of smart cities, and we like names exposed to the digital transformation of industries, such as domestic software solutions leader Yonyou and construction software leader Glodon; we also like vendors with direct exposure to smart cities, like cybersecurity leader VenusTech.

Information technology is changing the evolution of cities. Smart cessing, and access to information. Smart cities depend on sensors, cities run on IoT technologies, with data used to improve energy man- gateways, control centers and the cloud communicating efficiently agement, reduce environmental footprints, increase the safety of cit- through a mesh network with hierarchical elements. Each data gath- izens, and improve the maintenance of public infrastructure and ering, aggregation or decision point will have its own requirements buildings. A smart city needs an IoT network comprising MCUs, sen- that can only be met through flexible, scalable semiconductor solu- sors, communication ICs, and power ICs, with extensive software. The tions. total solution also comprises an integrated wireless sensor network platform that can monitor, control, and deliver smart city informa- For people to enjoy the full benefits of smart cities, some kind of com- tion and services. The platform can also be interfaced to motion, envi- munications system must connect all manner of smart things. ronmental sensing, and proximity sensor boards. Smartphones, tablets, and other portable devices today allow citizens to easily connect to many devices to get real-time data. Many New actors will be added to the IoT. Besides smartphones, all kinds IoT applications will not need to stress these new reliability require- of computers, vehicles, streets, buildings, and household appliances ments. Smartphones and many other consumer-facing connected will become part of a comprehensive communications infrastruc- devices are good enough, but other applications such as medical ture. The focus is no longer on direct communication between indi- devices, self-driving cars and many industrial/infrastructure IoT vidual people or devices but on linking countless users, devices and applications are mission-critical devices that require cutting-edge systems with each other. Roads, lights, parking lots, water supplies, technology to guarantee reliability. and a whole range of smart city applications need sensors, pro-

MORGAN STANLEY RESEARCH 71 M BLUEPAPER

Exhibit 108: Exhibit 109: Smart home devices expected to reach 3.6bn units by 2023 Smart built-in devices to reach 2.2bn units by 2023

Source: Canalys

Technology companies must understand the entire IoT stack’s Exhibit 110: requirements and, in some cases, create an end-to-end solution that IoT stack diagram facilitates market adoption. High-performance, high-bandwidth and Data Centers/ Cloud multichannel operation is critical, and scalable technology devices, (analytics, management and archive) software and solutions are increasingly important. A typical IoT stack Communication Security with a representative electronic subsystem consists of: Edge IT (analytics, pre-processing) l Sense. Typically the device that contains the sensors. It could Encryption Communication be a mobile phone or a medical device that captures a person’s heartbeat. IoT gateways l Analyze. The IoT platform also has the capability of inter- (data aggregation, A/D measurement, control) TPM preting data and sending analytics up through an application Communication that provides insights and alerts to people or providers. IoT devices l Communicate. Both the device and the IoT platform it commu- (Sensors and Actuators)

nicates with have the capability to analyze and organize the Source: Morgan Stanley Research data received. l Security. This is a critical component at every step of the journey, from the time the data is collected to the time it is served up in the application. l Cloud. This is a service outside the IoT stack, which includes the cloud platform, analytics/AI, and ease of cloud integration.

72 M BLUEPAPER Connected device demand explores when Exhibit 111: Connected devices in China: Shipments in 2018 vs. 2030 everything needs to be connected mn units 18,000 16,000 Connected device market value to more than double in 2030: 14,000 CAGR = 15% Increased connectivity among various devices allows for seamless 12,000 10,000 data transmission and enables AI innovation to enhance quality of life 8,000 in smart cities. In 2018, total connected device volume amounted to 6,000 CAGR = 9% 1.3bn units, among which infrastructure connections and smart- 4,000 phones were two major applications. We estimate total connected 2,000 0 device shipments will increase at a 9% CAGR to reach 3.9bn units by 2018 2030E 2018 2030E 2030, with a split of 65% public sector, 23% household and 12% per- Annual shipment Installed base Public Sector Household Personal sonal usage. Source: IDC, Morgan Stanley Research

In view of already saturated smartphone penetration currently and Exhibit 112: high value adds in public connected devices, we estimate total con- Connected devices in China: Value in 2018 vs. 2030 nected device value will grow to US$684bn in 2030, more than US$bn double the level of US$301bn in 2018, implying a 7% CAGR. The con- 3,000 tribution from public sector applications will also increase to 75% of 2,500 CAGR = 12% total value in 2030 vs. 49% in 2018. 2,000

1,500 Total installed connected device value to amount to US$2.7trn in 1,000 2030: We assume infrastructure and household connected devices CAGR = 7% 500 have replacement cycles of 4-5 years while personal devices are 0 replaced every 2 years. This implies that total installed connected 2018 2030E 2018 2030E devices will amount to 16.6bn units in 2030, up from 3.1bn units in Annual shipment value Installed base value Public Sector Household Personal 2018, or a 15% CAGR. Source: IDC, Morgan Stanley Research

We therefore estimate the total value of installed connected devices at US$2.7trn in 2030, more than triple the level of US$684bn in The ability of an IoT sensor or device to sense its surroundings, com- 2018. The substantial increase will be fueled by the rapid adoption of municate its state, and process collected data to determine the best connected devices in the public sphere (connected vehicles, infra- response to its environment require low-power consumption, high- structure connections, etc.) and households (lighting, home moni- speed precision, high-performance, integration and affordability. toring/security, smart speakers, etc.). – Sensors and micro-actuators sense and act. Sensors are going to Key technology requirements of smart play a critical role in collecting and processing data across a variety of industries. Sensors must perform for years without interruption or cities failure. This requires ultra low-power ICs with prolonged battery life through the use of energy harvested. They also vary in complexity, Smart cities will depend on the Internet of Things (IoT) applied combining multiple elements for temperature, sunlight, radar, lidar on a city-wide basis. The technology required for individual sensing or chemical detection, together with analog-to-digital conversion nodes in smart cities will be miniature and affordable, while con- (ADC) and signal amplification, local processing, a communications suming very little power. The metropolitan networks for power, interface, wireless transmission, a battery, and power management. communications, transportation, resource management, and other services will be integrated, and all of these areas will be served by Video cameras produce a large data stream, making bandwidth- and countless points of information collection feeding into the cloud. power-reduction techniques especially important. Cameras can wake up at intervals, or only when motion is detected. Frames can be scanned within the unit for objects of interest, allowing the selection

MORGAN STANLEY RESEARCH 73 M BLUEPAPER of only the most essential information for transmission. Advanced We expect demand from the government to grow fastest, with the compression technology will keep to a minimum the bandwidth installation base rising at a 20% sales CAGR during 2018-30. needed for communication. Object recognition and compression are Additional cameras will be installed to deter crime. AI cameras are enabled by video processors with acceleration for high-speed signal able to recognize faces and compare them against databases, which processing. can save hundreds of hours as compared with using manpower alone. Traffic monitoring systems combine surveillance cameras from train – High-performance processors, ultra-low power microcon- stations, airports, and main roads, and traffic data is analyzed in real trollers and security process information. The aggregation of time at the back-end server with AI enabled. Such systems can better transmitted data from multiple sensor and camera nodes is process- manage traffic and avoid traffic jams. For example, in Guangzhou, ing-intensive and requires high-performance computing to evaluate there are over 80,000 surveillance cameras as part of the traffic incoming data and decide what action should be taken locally. It is monitoring system, covering major roads and high-traffic areas. We more efficient, for example, for traffic control on part of an expect this will first be applied in urban areas and then penetrate into expressway to be processed at the edge rather than the main rural areas in China. Other than human and vehicle identification, gateway. these AI cameras are capable of floating object detection, sewage disposal detection, and wild animal detection. – Ultra-low power communication modules. Wireless communication plays a pivotal role in enabling a wide network of sensors, In commercial use, AI cameras can be applied in various sectors, such driving the need for stable technologies with low power require- as retail, manufacturing, education, healthcare, and energy. ments. As such, advanced CMOS technologies are needed to enable Surveillance companies such as Hikvision are cooperating with shop- low-power consumption and low cost. ping malls on pedestrian counting and behavior analysis, with precision manufacturing plants on flaw detection, and with schools on l Analog and mixed signal components translate the information distance education. In some scenic areas, cameras are helping people l Connectivity protocols, cloud providers, analytics and system to view scenery via VR. integrators l Power and energy-management modules keep the system run- We expect the professional surveillance camera installation base to ning in the most energy-efficient way. Efficiently managing reach 882mn by 2030, at a 17% CAGR during 2018-30. The number sleep cycling is critical to IoT implementation. of surveillance cameras will increase to 62 per 100 people by 2030 from 16 per 100 people in 2018, we estimate. This would be signifi- – Storage efficiency. Given the amount of data that the IoT is going cantly higher than the rest of the world, which we expect to have 17 to generate, storage requirements and costs should not increase cameras per 100 people by 2030. We expect penetration of exponentially, but rather data optimization and filtering should be AI-embedded cameras to reach 14% by 2030, up from less than 1% in used to limit storage. Software-defined storage can be used to bal- 2019. Growth will be supported by demand from the government ance and optimize the usage of available storage. (such as the Xueliang Project) and commercial applications such as unmanned supermarkets. – AI-embedded software platform for data management and cyber security. AI-embedded software will allow efficient data man- We also expect the consumer surveillance camera market to grow at agement in the integrated platform and, most importantly, provide a 30% sales CAGR during 2018-30. According to IDC, 9.7mn security effective feedback and/or predictions to assure sound operation in cameras were shipped to consumers in China in 2018, mainly for cities. Cyber security is also increasingly important to prevent attacks home security. We believe more households have demand for these and reduce unnecessary disruption. cameras, which can recognize strange faces, detect abnormal behaviors such as falling down, and monitor and record the important moments of babies or pets. IHS estimates that the number of house- Surveillance cameras in smart cities holds with cameras will increase to 67mn by 2030 from just 8.3mn in 2018, and camera installations per household will increase to 5.9 With AI embedded, surveillance cameras are becoming an important by 2030 from 2.3 in 2018. tool to improve efficiency and productivity in smart cities. We estimate the professional surveillance camera market will grow at an 8% sales CAGR during 2018-30, fuelled by new installations until 2025 and replacement demand thereafter.

74 M BLUEPAPER Exhibit 113: Trend toward localization creates new HIKVision: AI cameras can identify persons and their vehicles once facial features are analyzed at the back end opportunities for Greater China semiconductors

The potential global economic slowdown is likely to hit demand for tech products given their discretionary nature. We are therefore unsure whether logic semi inventory digestion will continue, thus making the cycle recovery slower than expected. Although we are starting to see some inventory depletion, the demand outlook is still uncertain.

Although the timing of a recovery in the logic semi cycle is unclear, we think the trend towards localization will continue, and offset potential order cuts in smartphone units in 4Q19. In May, we thought US semi components would run out in 2-3 months. However, demand Source: AI exhibition at HIKVision's office building, Morgan Stanley Research from Chinese system houses has sustained, given better allocation of critical component usage. Also, US semi companies have resumed Exhibit 114: some shipments of legacy components such as 4G FPGA for 4G base HIKVision’s facial recognition solution stations.

With trade tensions leading to uncertainty for the supply chain, we believe the push for China to localize its semiconductor industry will continue. We believe the design/R&D would be a way for China's semi industry to pursue an industry upgrade. And this is unfolding rapidly, with Huawei’s captive design house Hisilicon launching several projects in the Taiwan supply chain.

Increasing focus on efficiency and innovation drive long-term demand for software and IT services

We believe rising demand for software and outsourced IT services is being driven by Chinese enterprises' increasing focus on internal efficiency to combat slowing top-line growth and higher labor costs. Source: Company data, Morgan Stanley Research Further tailwinds are coming from technological innovation (in areas like artificial intelligence), previous underinvestment in software (historically low software spending as a % of GDP) and government policy support (top down digital transformation requirement in public sector and the general promotion of domestic technology in private sector). We expect China's overall software and IT services market to grow at a 15% CAGR through 2022, then at a 13% CAGR through 2030. We forecast software spending as a % of nominal GDP will more than double from 0.34% in 2018 to 0.76% in 2030. Cloud services, industrial-related software and cybersecurity are potential submarkets that could outgrow the industry average. MORGAN STANLEY RESEARCH 75 M BLUEPAPER Key challenges for technology companies Stock implications

TSMC IoT is a nascent and fragmented market with different end-customer needs. What the IoT industry seems to lack is consistent standards that enable interoperability and security of data. It is Within Greater China semiconductors, we think TSMC and MediaTek impossible for a single company to possess all the technologies or are best positioned. TSMC accounts for 50% of global logic chip pro- provide a blanket solution to various sub-segments of the IoT. As duction, and it is likely to be a key enabler of future 5G and AIoT appli- such, building an ecosystem of companies is critical, but striving to be cations. Its leadership should be sustained for at least another five the control point that manages the whole solution on behalf of eco- years, in our view, when Moore’s Law approaches to 3nm. system partners would be extremely challenging. Another challenge is on the trusted platform where numerous security gaps are created Will Semi during the integration and implementation of IoT systems. Will Semi is the #3 vendor of CMOS image sensors worldwide, with Making sense of the data. Data production rates from devices and 12% revenue market share in 2018, behind Sony and Samsung. As we machines are exponential, and processing and analyzing this data is move toward a more connected world, more and more electronic challenging. New approaches are required, such as search algo- devices will have to be empowered with machine vision functions to rithms, computation, visualization, and cloud-based processing. In communicate with each other. This increase in camera usage will addition, data privacy, data security and data quality research are inevitably translate into more sensor demand, allowing Will Semi to essential. address a larger market.

The coming limitations of Moore's Law will be a challenge for the GigaDevice development of cheaper and more powerful semiconductors that will enable the data era. Fundamentally, companies are not in the GigaDevice is the largest NOR Flash and MCU design house in China. business of continuously shrinking transistors but making attractive The company is the key beneficiary of IoT growth in the Chinese returns on investment and building useful products. But computing market, as both NOR and MCU are critical components in IoT devices. progress is becoming less predictable as we approach the limits of The company has also invested in D-RAM manufacturing, which is a Moore's Law, impacting performance, power consumption, and the key strategic move for Chinese semi localization. costs of computing. In addition, the required capital investments and operational costs of next-generation factories are likely to rapidly HIKVision exceed US$20bn within five years. Finally, the industry's ability to create chips that run faster while also using less power ended about In our view, HIKVision is well positioned in the video surveillance a decade ago. To continue advancing computing capabilities at market thanks to a combination of scale and the ability to offer com- reduced cost with economy-wide benefits will require entirely new prehensive customized solutions in smart city development. We semiconductor processes and device technologies. expect its revenue growth to remain at 20% YoY in 2019 and 2020 with sustained gross margin at 40%, thanks to China domestic Dependency on US technology and trade tensions. Dependence demand recovery. In the mid to long term, we anticipate additional on US technology in semiconductors is one of the key hurdles to sustainable revenue growth from: (1) AI enlarging the surveillance China's smart city development. For example, GPUs and FPGAs that industry market size; (2) expanding business into new machine vision are used in AI servers are still supplied by US semi companies. related fields, such as mobile robots/drones/auto cameras/industrial US-China trade tensions are forcing supply chain disruption, which is cameras; and (3) vertical integration of storage and AI chipset. detrimental to smart city development. Accelink Technologies High implementation costs. Implementation costs could be higher for AI servers and 5G infrastructure given the need to use leading- Accelink is one of the few domestic suppliers with in-house chip man- edge foundry processes. ufacturing capabilities. Ongoing demand for transmission components in the 5G era presents a bigger opportunity than does wireless business for Accelink; the datacom segment could present further

76 M BLUEPAPER upside. Both 5G infrastructure and data centers are developing along Glodon with urbanization in China. Our price target implies 46x 2019 and 34x 2020 earnings, based on our estimates. Examining historical pat- Glodon is a leading construction software vendor and a pioneer in the terns, we note that Accelink has traded at above 40x during upcycles, industry's digital transformation. With building information mod- i.e., 2014-16 (4G cycle) and 2009-10 (3G cycle). With the approaching eling (BIM) technology, Glodon's construction management product 5G upcycle, we believe that a valuation around 40x is justified. helps constructors to save material and labor costs, optimizes pro- Further, because of Accelink's in-house laser chip capability, we cesses to save time, and increases security in the construction stage. believe A-share semi stocks can also serve as a comparable valuation It can also help building owners to do predictive maintenance, secu- reference. rity checks and energy management. We expect the government's top-down encouragement of BIM adoption in smart cities as well as Foxconn Industrial Internet bottom-up constructors' increasing focus on efficiency to drive the long-term growth of Glodon's product. 8K+5G has been the key strategic development for Foxconn Industrial Internet (FII). Its full-range offerings, from 8K cameras, con- VenusTech tent creation (real-time raw data recording) and processing (cloud storage, network transmission) to data-driven analysis should allow VenusTech, as the biggest cybersecurity vendor in China, is well posi- it to win a number of smart city projects, such as those in Shanghai tioned to benefit, with the most comprehensive product portfolio, and Guangzhou, over the next 12-18 months. It is also leveraging part- especially for industrial internet and public infrastructure systems. nerships to enlarge its ecosystem, including strategic partners for VenusTech has established close relationships with high-profile cus- joint solution development (Advantech, Cognex, SAP), partners for tomers, such as governments and big SOEs, and built a strong brand integrated solutions (Yonyou, MegVII, CyberInsight) and service part- image in cybersecurity. Security operation service (SOS) is underpen- ners for local system integration and service support (Henry Waltz, etrated in China. The construction of smart cities could be an inflec- ViTex, IMRobotic). tion point for managed security services, due to the centralization of sensitive data and related security demand. VenusTech recently Yonyou issued convertible bonds and plans to use the proceeds to build four security operation centers. We expect managed security will drive Yonyou has been the largest enterprise software vendor in China and VenusTech's security service revenue at a 37% CAGR, 2018-21, with develops cloud services for further upgrades. It aims to leverage its managed security's revenue contribution rising from 17% in 2018 to cloud offering capability to create an integrated big data platform for 24% in 2021. smart city operation – open for multiple vendor solutions to enhance data management and analysis. Its AI-empowered software analytics can also help the city operate efficiently and safely. Policy support as an overseas vendor substitution fuels Yonyou's business potential in China.

MORGAN STANLEY RESEARCH 77 M BLUEPAPER Investment Theme #2: Digitalization of Old-Economy Industries

Exhibit 115: Investment Theme #2: Summary Key Beneficiary Key 2030 Forecasts Top Stocks

Early movers in EVs and autonomous EV Sales: Autos vehicles 8.4mn units (8x vs. 2018) • S.F. Holding (002352.SZ)

Logistics companies with strong R&D Express Volume: Logistics investment 300bn deliveries (6x vs. 2018) • NARI Technology (600406.SS)

Utilities and Power Utility players with competitive edge in Share of clean energy in capex: 60% Equipment smart grid (vs. 40% today) • Ping An Bank (000001.SZ)

Credit growth: 7% 2018-30 CAGR Banks More market-oriented banks (vs. 17% in the past decade) • Ping An Insurance Company (2318.HK) Insurers with leading positions in top- Insurance penetration: 9% Insurance tier cities and advanced technological (vs. 4.3% in 2018) capabilities • Yuan Longping High-tech Agricultural

Theme 2: Digitalization 2: Theme of old-economy industries GM corn and soybean seed application: Agribusiness entities with strong brand Agribusiness 50% (000998.SZ) name and GM seed pipeline (vs. 0% today)

Source: Morgan Stanley Research

78 M BLUEPAPER 2a. Autos

Overview: China's smart cities will drive three important trends in the auto segment: shared mobility, electric vehicles, and autonomous driving.

Key forecasts: By 2030 we expect 27.6mn passenger vehicles to be used in shared mobility in China, accounting for 10% of the total car parc vs. only 2% in 2018. We forecast that 33% of passenger vehicles sold in China in 2030 will be electric, vs. only 4% in 2018. We expect 20% of the PVs sold in 2030 to feature L4 or L5 levels of autonomous driving, compared with none in 2018.

Investment OEMs with sufficient R&D resources should benefit, such as SAIC, Dongfeng and GAC. In auto parts, early implications: movers into EVs/vehicle autonomy realms, like Huayu, Nexteer, Navinfo, should benefit from the growing electri- fication of fleets and the development of smart traffic that should hasten the adoption of autonomous driving.

New opportunities from China Auto 2.0 shared, electric and autonomous vehicles. Along with the growing impact of millennials, the auto value chain will be reshaped by a rising focus on environmental issues and energy efficiency, as well as rapid China continues to drive new mobility adoption: China is the transition into the new economy. world’s largest car market in terms of the number of new cars sold per year, and we think its use of electric and autonomous vehicles will Wealth created by China Auto 2.0 may surpass prior cycles: In the also lead the world in the next decade. This will profoundly influence new mobility cycle, we expect non-traditional automotive players, the pace of technological adoption globally. However, rather than such as autonomous driving solution providers, entertainment/ seeing traditional OEMs compete with tech giants, as in developed telematics content providers and shared mobility providers, will con- markets (Detroit vs. Silicon Valley), in China we see closer collabora- tribute up to 80% of China's market capitalization in the auto value tion between internet bellwethers like Baidu, Alibaba and Tencent chain by 2040. We believe the market is overly conservative in its (BAT), and traditional OEMs in developing electric and/or autono- view of new auto technologies, and while we acknowledge these mous vehicles in terms of hardware and software. We believe such changes are some way in the future, we believe the impact they will omni-channel collaboration will redefine the future of mobility in have is beginning to be felt now and is being reflected in investor per- China at a faster pace than in the rest of the world. ceptions, strategic actions (M&A, tie-ups), and stock valuations. In the future, we think car sales volume is likely to be a less relevant Major changes in auto technologies/operation will create new metric when evaluating the auto business; instead, we think innova- dynamics in China: We believe new auto market entrants will accel- tion, miles traveled and ARPU will become more important metrics erate their efforts to make inroads into this Auto 2.0 market of in boosting market capitalization.

MORGAN STANLEY RESEARCH 79 M BLUEPAPER

Exhibit 116: Exhibit 117: Reshaping the China auto value chain's capital allocation Potential industry value generated from Auto 2.0

Source: Morgan Stanley Research estimates

Outlook for China's electric vehicles of EV portfolios will likely take place in association with the suspension of investment in ICE. Migration could take more time but it's unlikely that OEMs and tier 1 players would sud- We remain optimistic on the long-term growth trajectory for denly turn back to ICE, given the lengthy lead time required for EVs: We stay constructive on EV demand over a three- to five-year new model development. Toyota officially targets 50% of its timeframe. We look for 1.2mn units of passenger EV sales for 2019 global sales to come from EVs by 2025, five years ahead of its (vs. consensus for 1.4-1.6mn units), but forecast a 19% CAGR over previous schedule. BMW announced it was speeding up the 2018-30, to 8.4mn units. That implies 33% market penetration by plan to electrify its model range by introducing 25 EVs by 2023, end-2030. two years ahead of its previous target. l Likely irreversible pursuit of electrified user experience: 30%+ Our positive stance mainly rests on the following reasons: of new car purchases are by millennials – a tech-savvy group of early EV adopters. The rise in the car-buying mix of the younger l Multiple regulatory levers for the Chinese government to pull: generation bodes well for the shift to EVs. We believe EVs remain a key strategic focus area for the government, with an eye toward global automotive dominance. The major pushbacks from investors regarding our EV forecasts l Strong model pipeline: Most if not all global OEMs will have include: more serious model launches from 2019 – for example, Mercedes-Benz EQC, BMW iX3 and Audi e-tron. Meanwhile, EV l Will there be real demand for EVs to back our bullish forecasts, startups and local brands are also vying to launch EVs, with as current demand is policy-driven? Although it is true that cur- over 50 new models scheduled for 2019-20. We believe the rent demand is mainly policy-driven, the cost of EVs has been profusion of new model launches should further increase EVs' falling at a rapid pace thanks to scale benefits and tech mindshare. upgrades. We look for EV demand to meaningfully take off l Huge sunk costs from investments in China, from both the gov- beyond 2022 when they reach cost parity with ICE vehicles. ernment and private sector: According to the Ministry of Meanwhile, quality upgrades will give EVs more traction with Industry and Information Technology (MIIT), the accumulated Chinese consumers, especially younger ones. investment in developing China's EV market topped Rmb2trn l Can the EV supply chain in China support such growth? We by January 2019. We believe this will hinder any potential believe China’s EV supply chain is competitive on all fronts, attempt to change course by either the auto companies or the from materials to components, given the country's early-mover government. advantage. China is also opening the market more aggressively l Major global OEMs and tier-one suppliers have brought for- to global players, which should enhance the development of ward their EV technology/product roadmaps: The development the industry through foreign capital, talent and technology.

80 M BLUEPAPER Outlook for China's Exhibit 118: We remain constructive on the long-term outlook for EVs in China – we forecast a 19% sales autonomous driving CAGR over 2018-30

We expect China's autonomous driving development to go through three stages:

China AD 1.0 (2018-21) – L1-3 autonomy – Competition driven: We expect L1-2 models to gradually penetrate the market and make up 25% of new car sales by 2020, as we notice OEMs are leveraging new features and smart car branding to boost sales. We expect to see incremental demand for radar/cameras, HD navigation maps, and car connectivity functions. OEMs and the supply chain, per our checks, now expect volume-manufactured L3 vehicles inte- Source: CAAM, Morgan Stanley Research estimates grating multiple ADAS functions to come after 2020. Exhibit 119: China AD 2.0 (2022-25) – L4 autonomy – Investment focus for different stages of China's autonomous driving (AD) development Tech driven: Parts/system makers we talked to expect L4 models to come to market as early as 2021 and achieve mid-single-digit penetration of new car sales by 2025. The key contents added will be execution functions (steering, braking, parking, acceleration) and algorithms (chips, MCUs), and applied in certain areas (e.g., logistics, shuttles).

China AD 2.0 (2026-30) – L5 autonomy –

Morality driven: Most of the industry con- Source: CAAM, Morgan Stanley Research estimates tacts we talked to expect L5 autonomous driving vehicles to hit the road in China after Exhibit 120: 2025, as the moral and ethical issues have Autonomous driving levels explained yet to be clearly defined. We think L5 vehicles will first be adopted by fleet runners for logistics and shared PVs, rather than by private users. Fleet operators would have a stronger incentive, i.e., large savings in labor costs, scale benefits from higher vehicle utilization, and additional revenue streams from a sizable road data pool. By 2030, we expect 20% of new vehicles sold to be equipped with L4 or L5 autonomous driving, vs. only 5% in 2025 and zero in 2018.

Source: SAE International, Morgan Stanley Research

MORGAN STANLEY RESEARCH 81 M BLUEPAPER Shared mobility and smart Exhibit 121: NavInfo provides lane-level road condition information and prediction traffic systems to solve congestion

Expect shared mobility to improve the utilization rate of vehicles: We forecast 10% of China's total car parc will be used in shared Source: NavInfo mobility in 2030, vs. only 2% in 2018. We expect a vehicle used in shared mobility could Exhibit 122: travel 5x more per year than a privately-owned Autostadt car towers vehicle, and the ride-sharing service could reduce the number of vehicles needed on the road, improving the efficiency of the traffic system.

Smart maps to improve the efficiency of road traffic: It has become a common practice for people living in cities to check road conditions before going out to avoid traffic congestion. NavInfo (002405.SZ) in China launched dynamic traffic information digital map products. The dynamic traffic information is developed and operated by Cennavi, a subsidiary of NavInfo. The company supplies products and services in 340+ cities across China as well as some countries and regions in Southeast Asia. It has over 4mn vehicle users and over 500mn Internet users, and more than 100 government Source: Autostadt and enterprise users. Exhibit 123: China's cities have insufficient parking slots Vertical parking lots to expand capacity for total vehicles: The Volkswagen Group has built two car towers at Autostadt in Wolfsburg, Germany. According to Volkswagen, each of the fully automated, high- rise stacks has space for up to 400 vehicles. The new cars are rolled over from the neigh- boring Volkswagen plant using a robotic-pallet system mounted on rails. The cars are loaded into and fetched from the towers using two 'car shuttles' or lifts per tower, each servicing 180° of the silo. We believe similar designs could serve as potential solution for the lack of parking space in major cities in China.

Source: Xinhua, People.com.cn, Morgan Stanley Research.*Beijing's data is residential only: residential parking slots divided by residential vehicles

82 M BLUEPAPER 2b. Logistics

Overview: Thanks to the development of city clusters and new technologies, we expect China's logistics costs to fall dra- matically. We expect 2030 express volumes to be five times higher than in 2018, with fulfilment times cut in half.

Key forecasts: – Logistics costs as a percentage of GDP: 14.8% in 2018 and 10% in 2030. – Express volumes: 300bn deliveries in 2030, up from 50bn in 2018. – Delivery times: 12 hours within city clusters and 24 hours nationwide.

Investment We like logistics companies that invest more in R&D, such as S.F. Holding and Deppon Logistics. implications:

Exhibit 124: A snapshot of technology applications in the logistics industry

Source: Deloitte

MORGAN STANLEY RESEARCH 83 M BLUEPAPER Lower costs Faster fulfilment

We expect social logistics costs to drop from 14.8% of GDP in 2018 Faster shipments: Shipment speeds can be increased two ways: to 10% in 2030, vs. 8% for the US in 2018. faster line-haul shipments and faster last-mile delivery. For line-haul, railways could gradually replace trucks in long-haul transportation. How to achieve lower costs The maximum speed of trucks is required to be no more than 120km/ h in China, while the speed of railways can reach 160km/h, 250km/h l Supply chain: Less administrative work; lower inventory levels; and even higher. For last mile-delivery, we believe unmanned drones, lower risks (i.e., insurance fees); lower damage rates which can easily reach low-density suburban areas, will provide low- l First mile: Optimization of cargo loading; higher utilization for cost instant delivery options. trucks; less administrative work; cost-saving from the use of unmanned drones Lower efficiency loss: Efficiency loss happens at certain stages, l Logistics hubs: Automation and robots replacing human labor such as (1) pick-up, (2) sorting, (3) line-haul, and (4) delivery. In each l Line-haul: Unmanned trucks; route and fuel-consumption opti- area, we expect efficiency to be improved by adopting new technolo- mization; remote control; 'road to rail' gies or hardware. l Last mile: Data analytics and dynamic planning; automation (i.e., smart lockers, robots and unmanned drones); electric vehi- l Pick-up/delivery: smart lockers, unmanned drones cles; less administrative work l Sorting: automation, robots l Line-haul: unmanned drones/trucks, high-speed rail to allow According to PwC, logistics costs could drop by 47% between 2018 shipment at night and 2030 worldwide. Moreover, with the help of big data, fulfilment times can be cut as optimized transport routes are adopted.

More efficient supply chain: In manufacturing & assembly, IoT and big data will enable predictive demand forecasts and real-time reac- tions to changes in demand and supply. This will reduce lead times and increase the utilization rate of warehouse space and other resources.

Exhibit 125: Exhibit 126: China: Social logistics costs as a % of GDP China: Breakdown of social logistics costs (2018) 14,000 24% Overhead 12,000 22% expenses 13% 10,000 20%

8,000 18%

6,000 16% Transportation 4,000 14% cost 2,000 12% 52%

- 10%

1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Storage cost China: Logistics cost (Rmb bn) as % of GDP 35% Source: CEIC

Source: NDRC

84 M BLUEPAPER

Exhibit 127: Case study: How a parcel gets from Shanghai to Nanjing 3km 30km 300km 30km 20km

Merchant Shanghai Nanjing Customer 23 hours Local Outlet 2 hours 6 hours 3.5 hours Local Outlet 1.5 hours (Shanghai) Sorting hub Sorting hub (Nanjing)

13 hours Source: Morgan Stanley Research. Note: Parcel delivered in August 2019

For express parcels, by 2030, we expect fulfilment times to fall to 12 Exhibit 128: hours within city clusters (vs. 24 hours in the Yangtze River Delta Express parcel volume per capita (2018 vs. 2013) today) and 24 hours nationwide (vs. 2-4 days today). 200

160 In September 2019, Cainiao, together with ZTO, Yunda, YTO, and other express companies, announced 24-hour fulfilment for parcel 120 shipments that start and end in 26 cities in the Yangtze River Delta. 80

40 Larger volumes - 2013 2018 We expect China's express deliveries to grow at a 16% CAGR in Fujian Jiangsu Beijing Guangdong Shanghai Zhejiang 2019-30, to 300bn per year, driven by increasing disposable income Source: CEIC per capita and the e-commerce boom. We expect parcels per capita to reach 365 annually in major city clusters and 110 annually outside Exhibit 129: these areas. In 2017, Jack Ma mentioned at a smart logistics summit 2030 population in major city clusters and other areas that he thinks annual express volumes in China will reach 365bn by Remaining Admin Areas 2025 (Securities Daily, May 24, 2017). This is more aggressive than 7% our forecasts.

Major Clusters 41%

Other Cities 52%

Source: Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 85 M BLUEPAPER Stock implications

We prefer companies with higher R&D and IT spending. We expect the application of new technologies and hardware to significantly help players to lower costs, improve efficiency/service quality, leading to competitive advantages and higher pricing power compared with their peers. We also expect innovations in logistics to help reshape traditional business flows and cash flows in the economy, leading to new business areas for logistics players.

Exhibit 130: Exhibit 131: China's express firms: R&D/IT spending as a percentage of total sales Absolute spending on R&D/IT in 2018 (Rmb mn) 3.0 2,500

2.5 2,000

2.0 1,500

1.5 1,000

1.0 500

0.5 - SF STO 0.0 YTO Yunda SF Deppon STO YTO

Yunda FY18 1H19 Deppon Source: Company data. Note: S. F.: R&D expensed and capitalized; Deppon: IT spending FY18 1H19 Source: Company data. Note: S. F.: R&D expensed and capitalized; Deppon: IT spending

Case study: S.F. Holding

Exhibit 132: S.F. Holding: Application of technologies in logistics

Source: Company data

86 M BLUEPAPER Case Study: Deppon Logistics

Exhibit 133: Deppon: Application of technologies in logistics Digitalization Improves Management

Volume Di Lu Full Process Intelligent GIS Smart Sorting Hub Digital Twin Finding Cargo Forecasting and System Visualization Services Load Balancing Intelligent Services Improve Customer Experience

Intelligent Package Electronic Multi-Functional Collection and Smart Voice Waybill Handset Delivery New Technology Hardware improvement

Mixed Parcel Second Deppon Self-driving Sorting Generation PDA D Plus Cars

Source: Company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 87 M BLUEPAPER 2c. Utilities and Power Equipment

Overview: Given the Chinese government’s commitment to fighting climate change, we expect China to continue to increase clean energy (mainly nuclear, hydropower, wind, and solar) capacity. By 2030, the grid system will involve a complex distribution network to integrate distributed renewable energy and electric vehicles. Also, the grid system will need to embrace more renewable energy (mainly wind and solar). A smart grid is needed to ensure the safety and reliability of the power system.

Key forecasts: – The power generation capacity of non-fossil fuels will increase to 60% by 2030 from 40% in 2018, driven by solar, wind and nuclear, per our forecasts. As a result, power generation from non-fossil fuel sources will reach 37.5% in 2030 from 29.8% in 2018.

– We forecast that China’s investment in the smart grid will increase 2.6x, to US$80bn during 2021-30 vs. US$30bn over 2011-20.

Investment We expect companies with exposure to nuclear power plant operations, such as CGN Power, and those exposed implications: to smart grid equipment manufacturing, such as NARI Tech, will benefit.

High targets for non-fossil fuel generation tion in 2018 was 6,990bn kWh, of which non-fossil fuel generation accounted for 30.9%, contributed mainly by hydropower, wind, by 2030 nuclear, and solar at 17.9%, 5.2%, 4.2%, and 2.5%, respectively.

The Chinese government has committed to reducing the proportion We expect China to continue to contain new thermal power capacity of fossil fuel consumption, and aims to further increase non-fossil during 2021-30, and add more wind and solar capacity, which are fuel generation to 50% in 2030, according to 'The revolution more economical and flexible and also less controversial than devel- strategy of energy production and consumption (2016-2030)' oping hydropower and nuclear power. We expect acceleration of announced in December 2016. China is well on track to accomplish nuclear approvals in 2020 upon commissioning of China's first its goal of non-fossil fuel generation of 30% in 2020, compared to Hualong One reactor. 26.9% in 2015 and 19.2% in 2010. The power generation capacity of non-fossil fuels will increase to China's total power capacity reached 1,900GW, including thermal 1,856GW or 60% of total capacity by 2030 from 907GW or 40% of power of 1,147GW (60.4%), hydropower of 350GW (18.4%), wind total in 2018, driven by solar (26.9%), wind (14.8%) and nuclear power of 184GW (9.7%), solar power of 175GW (9.2%), and nuclear (3.7%), based on our forecasts. As a result, power generation from power of 45GW (2.3%) as of the end of 2018. Total electricity genera- non-fossil fuel sources will reach 37.5% in 2030 from 29.8% in 2018.

88 M BLUEPAPER Exhibit 134: Exhibit 135: Cumulative capacity breakdown Power generation breakdown (GW) (TWh) 3,500 12,000 Solar Nuclear Solar Nuclear 3,000 Wind Hydropower Wind Hydropower 10,000 Thermal Thermal 2,500 8,000 2,000 6,000 1,500 4,000 1,000

500 2,000

- - 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E 2019E 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E

Source: China Electricity Council, Morgan Stanley Research estimates Source: China Electricity Council, Morgan Stanley Research estimates

Future grid system: Embracing renewable Besides greater fluctuations on the supply side, a new form of energy demand – electric vehicles – is been emerging. EV charging infra- energy plus complex distribution networks structure has experienced a high double-digit CAGR in the past five years, and is likely to maintain high growth in the future, driven by China will continue to connect to renewable energy, and we forecast rising demand from EVs in light of advances in technology, changing its capacity contribution will rise to 23.3% in 2020 and further to economics, and climate concerns. Increasing EV penetration will 33.5% by end-2025. Besides the scalable renewable projects trans- raise the need to reconfigure distribution networks to alleviate con- ported via transmission networks across vast distances, the govern- gestion caused by EV charging demand. Energy demand will become ment is also encouraging renewable energy to be generated locally more dynamic and complex. and connected directly to distribution networks.

Exhibit 136: The future of the power grid system: Embracing more renewable energy for transmission involves a complex distribution network integrating distributed renewable energy and electric vehicles

Smart dispatch system

High voltage to Medium voltage to low High voltage medium voltage voltage secondary step up primary substation substation substation

Storage Power flow Information flow

Source: Morgan Stanley Research; Note: Red icon stands for new elements in the grid system

MORGAN STANLEY RESEARCH 89 M BLUEPAPER Exhibit 137: Stock implications Growing overall smart grid/secondary equipment demand/penetration driven by grid system evolving NARI Technology Secondary equipment invesment/total equipment investment 25% 25% Stable competitive landscape in high-end smart grid/secondary equipment market: Given the nature of secondary equipment for 20% 20% ensuring the operational safety of the grid network and the reliability 15% of electricity supply, grid companies (State Grid and China Southern 11% Grid) only have a short list of qualified domestic suppliers, especially 10% 8% for high-end equipment involving control and monitoring functions, despite conducting a open tender for equipment procurement. 5% 4% Secondary equipment includes software-oriented or software and

0% hardware integrated products, thus pricing and gross margins 2006-10 2011-15 2016-20e 2021-25e 2026-30e (>20%) are relatively stable. When adopting new functions, the ASPs Source: China Electricity Council, Morgan Stanley Research estimates of upgraded products are likely to be increased. The competitive landscape in the high-end secondary equipment market remains Exhibit 138: stable. There are a few players, including NARI, XJ Electric, Sifang Spending on smart grid/secondary equipment to increase by 2.64x Automation, Sieyuan Electric, Changyuan Group, and Guodian during 2021-30 vs. 2001-20 Nanjing Automation. (Rmb bn) 600 568 Leading market position backed by strong R&D: NARI (ultimate 500 parentco, State Grid Electric Power Research Institute) is actually the 400 2.64x R&D center of State Grid for high-end secondary equipment, and has 300 attracted, cultivated, and retained a group of talented experts in the 215 200 grid system. NARI has replaced international players in the past decade, to become the only supplier of a grid dispatch platform at 100 state and provincial levels. NARI also supplies the state level dispatch - system for the Philippines. The company has a higher market share 2011-2020e 2021-2030e in some other high-end secondary equipment, such as high voltage Source: China Electricity Council, Morgan Stanley Research estimates level substations (>35%), high voltage level relay protection equip- Grid spending to remain high over 2021- ment (>40%), as well as master stations (50%). 25, requiring more smart grid/secondary CGN Power equipment As the Chinese government phases out coal power, the only remaining power sources available will be gas, nuclear, hydropower, Smart grid/secondary equipment refers to software-oriented prod- biomass, wind, and solar. Comparing various power sources (with the ucts that control, regulate, protect, and monitor the grid system. exception of hydropower), we believe nuclear is competitive vs. coal With the evolution of both transmission and distribution networks, as a power source for base-load, as well as vs. other clean and renew- we expect the overall secondary equipment investment penetration able power sources including gas, wind and solar. Moreover, among in transmission and distribution networks will increase to 20% all these sources, nuclear is the most cost-competitive at during 2021-25 and 25% during 2026-30, up from 11% in 2016-20, Rmb430-435/MWh, and it provides a high level of energy security. from both incremental and replacement demand. The replacement cycle of secondary equipment is normally 5-10 years. As a result, we China's nuclear power development was halted during 2016-18, with expect demand for smart grid/secondary equipment will see high three years of no new approvals, until 2019 when the government growth of 2.6x (to US$80bn) over 2021-30 compared US$30bn in approved three projects – Shandong Rongcheng CAP1400, Fujian 2011-20. Zhangzhou and Guangdong Talpingling, both Hualong One units. We expect an acceleration of nuclear approvals in 2020 upon the commissioning of China's first Hualong One reactor.

90 M BLUEPAPER 2d. Banks

Overview: Urbanization 2.0 should support consistent credit demand from infrastructure projects, industrial upgrading, and consumers. Smart cities and advanced technologies are likely to enhance banks' efficiency and risk management capabilities, expand the retail client base, improve capital allocation, and help lower financing costs to the real economy.

Key forecasts: Household financial leverage will rise by around 10ppts through 2030, to 56% of GDP, similar to the pace in Japan in the 1960s and early 1970s when the urbanization rate was approaching 70%. Overall system leverage will rise to around 300% of GDP by 2030 from 275% in 2018, on healthy total credit growth of 7-8% annually (vs. 17% in the past decade), per our projections.

Investment We expect market-oriented banks to be the key beneficiaries of Urbanization 2.0, with Industrial, PAB and CMB implications: remaining our top picks.

l Enablers of Urbanization 2.0 In addition, as an asset-heavy sector, infrastructure financing requires much capital. Currently around 80% of total system credit comes from the banking system in China, indicating China's banks have in-depth experience and sufficient capital to banks have sufficient capacity to continue to play a major role finance and facilitate infrastructure projects, in our view: in infrastructure financing. l Infrastructure accounts for around 30% of total credit outstanding in China and requires mostly debt financing, where banks have expertise and long experience. The relationships built between banks and local governments and SOEs in terms of infrastructure cooperation also help, providing a stable funding source for projects and better risk control for banks.

MORGAN STANLEY RESEARCH 91 M BLUEPAPER

Exhibit 139: Infrastructure accounts for around 30% of total credit outstanding in China Amount (Rmb bn) 2016 2017 2018 Manufacturing 20,187 20,080 20,603 Wholesale retail trade 14,083 14,780 14,306 Real estate 17,639 20,529 23,326 Construction (excluding infrastructure related) 11,165 11,923 12,243 Mining 5,166 4,905 4,858 Farming, forestry, animal husbandry & fishery 1,947 1,801 1,694 Overseas 3,255 3,431 3,602 Other 9,450 7,935 7,177 Corporate Credit (excluding infrastructure related) 82,891 85,384 87,809

Housing mortgage loan 18,055 22,139 26,499 Other retail 4,282 6,055 6,663 Retail commercial credit 8,168 9,114 10,450 Credit card 4,075 5,673 6,915 Automobile purchasing loan 792 942 1,077 Retail credit 35,372 43,923 51,604

Transport,storage&postal service 16,389 17,529 19,418 Water conservancy, environment & public utility mgt 14,069 15,241 15,265 Electricity, gas & water production & supply 8,010 8,402 9,235 Leasing & commercial service 15,757 19,951 21,937 Construction (infrastructure related) 6,422 6,846 6,442 Other basic services (IT, education etc) 1,952 2,083 2,033 Infrastructure credit 62,599 70,052 74,332

Central government bonds 11,988 13,434 14,880 Local government bonds 10,625 14,745 18,070 Bank restructuring debt 810 810 810 Government credit 23,423 28,990 33,761

Total credit to real economy 204,285 228,348 247,505 GDP 74,359 82,712 89,577 Total real economy leverage 275% 276% 276%

Infra credit % total 31% 31% 30% Infra credit (incl local government bonds) % total 36% 37% 37%

Source: PBOC, CEIC, Wind, China Trust Association, 01Caijing, P2P Eye, WDZJ, Morgan Stanley Research

92 M BLUEPAPER A beneficiary of more support for credit resources that could support interest payments in addition to tax revenue. On the other hand, FAI growth has historically been highly demand, enhanced efficiency, risk correlated with FAI funding growth. As a result, we believe infrastruc- management, and capital allocation, ture credit demand going forward will support credit growth, which will be largely in line with nominal GDP growth. enabled by advanced technologies However, we expect future infrastructure financing to be carried out More support for credit demand from infrastructure via loans, bonds (including local government bonds), and ABS rather development and industry upgrading than, as previously, through non-standard credit assets, as China's financial cleanup puts stricter requirements on asset/liability dura- We see still significant growth potential for infrastructure, and the tion matching, the gradual removal of implicit guarantees, and look- interest burden for local governments is still manageable as a whole through risk control for underlying assets. (for details see How market misperceptions undervalue China's banks). The infrastructure interest burden is running at around 1.61% In addition, industry upgrades in smart manufacturing and smart agri- of GDP in China in 2019, compared with 1.59% for the US and 1.64% culture will be capital-intensive. This will be another source of credit for Japan in 2018, and China's government controls more assets and demand supporting banks' asset growth and profitability, in our view.

Exhibit 140: Exhibit 141: Infrastructure interest burden accounts for around 1.61% for China in FAI growth has historically been highly correlated with FAI funding 2019, vs. 1.59% for the US and 1.64% for Japan in 2018 growth Interest burden as % of GDP FAI (RHS) FAI funding (RHS) bn Rmb 1.72% FAI yoy FAI funding yoy 1.70% 1.70% 45.0% 70,000 1.68% 40.0% 60,000 1.66% 35.0% 1.64% 50,000 1.64% 30.0% 1.62% 1.61% 25.0% 40,000 1.60% 1.59% 20.0% 30,000 1.58% 15.0% 20,000 1.56% 10.0% 1.54% 5.0% 10,000 1.52% 0.0% - China 2018 China 2019E US 2018 Japan 2018

Source: Wind, CEIC, Morgan Stanley Research

Source: CEIC, Morgan Stanley Research

MORGAN STANLEY RESEARCH 93 M BLUEPAPER More support for credit demand from Chinese households

We see potential for steadily rising household credit demand alongside improving household financial assets and higher disposable income, facilitated by Urbanization 2.0. We forecast that household financial leverage will rise to around 56% in 2030 from 46% in 2018, similar to the pace of household leverage increase in Japan in the 1960s and early 1970s as the urbanization rate was approaching 70%. We forecast household consumption credit growth of 8.1% annually through 2030. We expect the overall system leverage ratio (total credit as a percentage of GDP) to rise to 300% in 2030 vs. 275% in 2018, implying credit growth at a 7.1% CAGR (vs. 17% in the past decade).

Exhibit 142: Exhibit 143: With improving household financial assets and higher disposable Japan experienced a similar increase in household leverage as its incomes, facilitated by Urbanization 2.0, we see potential for steadily urbanization ratio rose to 70-75% in the 1960s and early 1970s rising household credit demand % Japan urban population % total Japan HH loan % total (RHS) 95 82%

90 72% 62% 85 52% 80 42% 75 32% 70 22% 65 12% 60 2%

Source: CEIC, Wind, Morgan Stanley Research

Source: Federal Reserve, Bank of Japan, Bank of Korea, Wind, CEIC, Morgan Stanley Research. Data as of 2018.

Exhibit 144: We forecast household financial leverage to rise to around 56% in 2030, with the overall system leverage ratio (total credit as a percentage of GDP) rising to 300% in 2030 2016 2017 2018 2030E Total credit to real economy 204,285 228,348 247,505 562,247 GDP 74,006 82,075 90,031 187,416 Total real economy leverage 276% 278% 275% 300%

Retail credit % total 13% 15% 17% 26% Household consumption credit % GDP 37% 42% 46% 56%

Source: PBOC, CEIC, Wind, China Trust Association, 01Caijing, P2P Eye, WDZJ, Morgan Stanley Research

94 M BLUEPAPER Advanced technologies and smart cities to help will make trade related information, product and cash flow more improve client base, efficiency, risk management and transparent and traceable, which should help reduce trade finance capital allocation credit risks, a key source of NPL, with a cumulative 30% NPL on trade finance credits digested in the past four years. The Urbanization 2.0 process will also help expand banks' target client bases with improved efficiency. On one hand, banks still More efficient and effective capital allocation enabled by largely target more prime customers because of their risk-averse advanced technologies. Before 2017 and amid Urbanization 1.0, nature and lower loan yield offered as compared with alternative China's fast growth was, to a large extent, fueled by excessive credit financing channels. Financially healthier households amid the urban- growth. This has resulted in a waste and misallocation of credit ization process should help enlarge banks' borrower candidate pool. resources and oversupply issues. With the help of more data on busi- On the other hand, advanced technologies should help enhance nesses, the economy and financial markets, we expect improved cap- banks' efficiency with existing manpower in terms of screening and ital allocation amid Urbanization 2.0, with more effective use of risk control, thus enlarging their client bases. credit to support the real economy.

We also expect improved risk management capabilities at banks Lower operating expenses and credit costs, as well as more efficient with reduced NPL risks. Smart cities connect households' and cor- and effective capital allocation could help create a wider profit buffer porates' data points to arrive at a more comprehensive borrower pro- for banks when guided to lower financing cost to better serve the real file via advanced technologies such as 5G, AI, and big data analysis. economy. Support on credit demand will also help with banks' asset This should help improve banks' proficiency in risk assessment and expansion and revenue generation. credit pricing. In addition, smart city and manufacturing technologies

Exhibit 145: Exhibit 146: Before 2017 and amid Urbanization 1.0, China's fast growth was, to a ...this has resulted in waste and misallocation of credit resources and large extent, fueled by excessive credit growth... oversupply issues Nominal GDP yoy Reported credit yoy Total credit yoy 70 Year-over-year change, % 20% 18.0% Nominal GDP 60 FAI of industrial industry 16% 50 Lagging 12.4% industrial 12% 10.9% 40 supply vs. 9.5% demand 10.0% 30 8% 9.3% 8.4% 20 4% 10

0% 0 2019E 3/2015 6/2015 9/2015 3/2016 6/2016 9/2016 3/2017 6/2017 9/2017 3/2018 6/2018 9/2018 3/2019 6/2019 Jun-03 Jun-06 Jun-09 Jun-12 Jun-15 Jun-18 Mar-04 Mar-07 Mar-10 Mar-13 Mar-16 Mar-19 Dec-04 Dec-07 Dec-10 Dec-13 Dec-16 Sep-05 Sep-08 Sep-11 Sep-14 Sep-17 12/2014 12/2015 12/2016 12/2017 12/2018

Source: PBOC, CEIC, Wind, China Trust Association, 01Caijing, P2P Eye, WDZJ, Morgan Stanley Research. Source: CEIC, Morgan Stanley Research E = Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 95 M BLUEPAPER l Stock implications PAB has a strong retail business, while it is also developing its corporate business and should benefit from Urbanization 2.0. On one hand, PAB targets mostly well-off clients, a segment We expect market-oriented banks to be the key beneficiary of which Urbanization 2.0 can help expand. On the other hand, Urbanization 2.0. Notably, Industrial, PAB and CMB remain our top PAB guided that a major focus for its corporate business and picks for the Urbanization 2.0 theme. WMP underlying assets is infrastructure, an arena in which Urbanization 2.0 should help generate sufficient high-quality l Industrial Bank has traditionally been strong in asset sourcing, assets. especially in the areas of property and infrastructure. The bank l CMB is well known for its strong retail business. The potential also has good relationships with developers, local governments for more household credit demand should help support CMB's and SOEs. It is also strong in bond underwriting. Urbanization loan growth, expand its client base, and contribute to the 2.0 should create more credit demand, while the development bottom line. of capital markets should increase bond underwriting demand and more standardized asset management products. Industrial Bank, in our view, should be a key beneficiary in terms of gaining infra credit and bond underwriting market share. Industrial Bank is also developing its retail arm, with some improvement in retail funding and loan extension, which should also help take advantage of more credit demand.

96 M BLUEPAPER 2e. Insurance

Overview: Continued urbanization should lift distribution efficiency and reduce claims, while new opportunities are created in agricultural insurance.

Key forecasts: China's total insurance penetration ratio will rise to 9% in 2030 from 4.3% in 2018, with total insurance premiums reaching US$2.3trn, implying a CAGR of 13%.

Investment Insurers with leading positions in top-tier cities and advanced technological capabilities, such as Ping An, should implications: benefit the most. Insurers with dominant positions and strong expertise in agricultural insurance, such as PICC P&C, should also benefit.

City clusters can improve operational and distribution efficiency: Chinese insurers are already amongst the largest globally in life, non-life and reinsurance. However, their expense ratios are not yet the lowest. Hong Kong insurers tend to have higher operational efficiency than their Chinese peers given that they operate in a compact city. AIA HK's G&A ratio is half that of Chinese peers despite gross premiums being 80% smaller. Its distribution efficiency is also higher, partly because of a high concentration of wealthy people and better city connectivity. As city clusters grow in China, the efficiency gap with Hong Kong insurers should start to narrow. We note that domestic insurers are already catching up by refocusing on metropolitan areas, recruiting better educated agents and utilizing digital tools to improve efficiency.

Exhibit 147: Exhibit 148: Hong Kong vs. Mainland insurers: G&A ratio* Hong Kong vs. Mainland insurers: Agent productivity* % Rmb

10.0 8.8 120,000 9.0 98,278 100,000 8.0 7.0 7.0 80,000 59,982 6.0 60,000 5.0 4.1 4.0 3.5 40,000 3.0 20,000 7,771 4,392 2.0 1.0 .0 PCA AIA AIA HK China life AIA HK

Case Ping AnLife

China life productivity HSBC HK* 1.70 1.13 1.22 1.00 Ping AnLife

*HSBC data as of FY15, others as of FY18 *First year premium per agent per month, as of FY18 Source: Company data, Morgan Stanley Research Source: HKIA, company data, Pi Fsi, Morgan Stanley Research

Exhibit 149: China: Life insurance market projections

Rmb bn 2014 2015 2016 2017 2018 2019E 2020E 2021E … 2030E FYP 657 931 1,388 1,536 1,163 1,324 1,464 1,625 2,674 Renewals 612 654 782 1,068 1,463 1,726 2,159 2,740 10,331 Total life premium 1,269 1,586 2,169 2,604 2,626 3,050 3,623 4,364 13,004 Growth rate % 25 37 20 1 16 19 20 13 Penetration % 2.0 2.3 2.9 3.2 3.0 3.2 3.5 3.9 6.9

Source: CBIRC, Bureau of Statistics, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 97 M BLUEPAPER Smart cities can improve profitability by reducing fraud and Aside from being key beneficiaries, insurers with advanced techno- claims risks: China's auto insurance claims frequency has been run- logical capabilities, such as Ping An, can also serve as enablers of ning at over 20% – much higher than the 5-6% in developed econo- China's smart city initiatives. Ping An has already built up its Smart mies. This is partially a consequence of lower road safety awareness City ecosystem by exporting core technologies to help local govern- and inefficient traffic management systems. With smart city initia- ments provide better citizen services and manage local economies. tives, most cities have now installed AI detectors in major streets and Some of Ping An's key platforms under its Smart City ecosystem are adopted more advanced traffic management systems to monitor listed in Exhibit 153 . The company has already secured smart city real-time violations and ease traffic congestion. We believe these ini- projects in more than 100 cities, and it is still looking to expand the tiatives will also help to reduce China's auto claims ratio. Ongoing scale of this business. digitalization and wider adoption of big data analytics are also helping insurers reduce fraudulent claims, which should help P&C insurers improve their profitability. We are expecting a potential 20ppt reduction in CoR due to technology advances in the future ( Exhibit 151 ).

Exhibit 150: Exhibit 151: China vs. US: Car damage claims frequency Technologies can further reduce the combined ratio by 17-21ppts % China US

120 100 100 89

80 61 51 53 51 60 46 41 40 30 28 23 20 5.4 5.5 5.7 5.8 5.6 5.7 5.9 6.0 6.1 6.1 6.1 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: Annual reports from European insurers, interviews, BCG Insurance Benchmark Database, BCG Source: Insurance Association of China, CBIRC, III, Morgan Stanley Research case experience, BCG Analysis, Morgan Stanley Research

98 M BLUEPAPER

Exhibit 152: Ping An's Smart City Initiatives

"1+N" platforms covering sectors under 3 concepts Government services Economic development People'sPeople's livlivelihoodelihood Smart Government Smart Customs Smart Life Smart Fiscal Smart Agriculture Smart Elderly Care Smart Community Smart Trade Smart Transportation Smart Urban Management Smart Finance Smart Poverty Smart Legal Smart Development Alleviation Smart Environmental & Reform Smart Real Estate Protection Smart Economy Smart Education Smart Market & Trade Smart Healthcare Regulation Smart Enterprise

Smart City Cloud

Source: Company data, Morgan Stanley Research

Exhibit 153: Key platforms under Ping An's Smart City Initiatives Module Platform Description Impact Smart An AI and blockchain-empowered business credit – Covers over 3mn businesses Government business platform that helps the government increase efficiency – 90% accuracy rate in identifying enterprises with services development in business administration abnormal credit profiles – Facilitated 100% full-chain traceability for hog A blockchain-based traceability system for the full hog raising, slaughtering and the pork supply chain Economic Smart production cycle that helps farmers improve the quality – Reduced producers' financing costs by 20% development agriculture of agricultural products – Gave consumers access to 100% food safety insurance coverage

An AI-based platform that provides smart traffic – Partnered with over 20 cities People's Smart traffic solutions such as AI supervision of traffic law violations – Traffic accident rates dropped 10% livelihood management and AI traffic predictions to ease congestion problems – Traffic jam times on key roads fell 30%

Source: Company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 99 M BLUEPAPER New opportunities in agricultural insurance: Despite an 11-fold Large insurers, such as PICC P&C, have already deployed advanced increase in agricultural insurance premiums thanks to strong policy technologies in this line to make risks more quantifiable. For example, and fiscal support over the past decade, China's agricultural insur- the company is using drones to do image-based underwriting and ance protection level remains low as compared with the US level claims assessment for farms in remote areas and is adopting facial ( Exhibit 154 ), partly as a result of a lack of data and effective claims recognition of pigs to prevent fraudulent claims. We forecast that assessment tools stemming from the country's fragmented and agricultural premiums will grow at a 17% CAGR to 2030. small-scale farming model. Wider adoption of smart farming and continued land reforms to facilitate large-scale farming could help Overall, we expect China's total insurance penetration ratio to rise to insurers better underwrite and control risks in agricultural insurance 9% in 2030 (vs 4.3% in 2018), within which life penetration should as well as expand coverage to a wider variety of agricultural prod- reach 7% and non-life should reach 2%. ucts.

Exhibit 154: Exhibit 155: China vs. US: Agricultural production and premiums Technology adoption in agricultural insurance Agricultural gross Agricultural premium Product Technology Change in Business USD bn output value USD bn 1,800 1,695 12 Image-based underwriting/ Crop insurance Drone/Remote Sensing Claim settlement 1,600 9.9 10 Digital risk identification system 1,400 8.5 Crop insurance Big data for catastrophe prevention 1,200 8 AI/Facial recognition for Livestock insurance Anti-fraud 1,000 livestock 6 800 Source: Company data, Morgan Stanley Research 600 445 4 400 2 200 0 0 China 1US China 2 US Source: CEIC, A.M. Best, China Insurance Yearbook, Morgan Stanley Research

Exhibit 156: China's agricultural and total P&C insurance market projections

Rmb bn 2014 2015 2016 2017 2018 2019E 2020E 2021E … 2030E Agricultural gross output value 9,782 10,189 10,648 10,933 11,358 11,812 12,285 12,776 16,670 Gross output value growth (%) 4.2 4.5 2.7 3.9 4.0 4.0 4.0 3.0 Protection level %* 15.3 19.2 20.3 25.5 30.5 34.5 38.5 42.5 75.0 Agricultural sum assured 1,498 1,960 2,160 2,785 3,460 4,071 4,725 5,425 12,502 Premium rate % 2.2 1.9 1.9 1.7 1.7 1.7 1.7 1.7 3.0 Agricultural premium 33 37 42 48 57 67 78 90 375 Growth rate % 15 11 15 20 18 16 15 17 Total P&C premium 754 842 927 1,054 1,176 1,305 1,467 1,647 3,844 Growth rate % 12 10 14 12 11 12 12 10 Penetration ratio% 1.18 1.23 1.25 1.28 1.31 1.35 1.40 1.46 2.05

*Protection level = sum assured/agricultural gross output value Source: Bureau of Statistics, CBIRC, Morgan Stanley Research

100 M BLUEPAPER Stock implications

Ping An PICC P&C

Ping An ranks #1 in terms of life market share in Beijing, Shanghai, PICC P&C is the dominant player in China's agricultural insurance Guangzhou, and Shenzhen, owing to its focus on higher-tier cities, market with a 50% market share. This is enabled by its large amount better-educated agents, and higher productivity empowered by of data, deep relationships with local governments, and strong advanced digital tools. As large city clusters grow, we believe Ping An underwriting expertise. We believe the modernization of agriculture will be a key beneficiary as it continues to leverage its technologies could present new opportunities for the company, as the wider adop- to improve distribution and operational efficiency and solidify its tion of smart farming and the transition to large-scale farming make leading position in large cities. Its smart city initiatives, which help insurance coverage more viable and available to farmers at an afford- local governments build smart city platforms, could also create new able price. The company should also benefit the most from smart city revenue streams for the group. development and see secular improvements in profitability.

MORGAN STANLEY RESEARCH 101 M BLUEPAPER 2f. Agribusiness

Overview: As farmland is consolidated, we believe demand for genetically modified seed will be boosted as it can reduce labor inputs, achieve higher yields and improve the economics of farming.

Key forecasts: We expect GM corn and soybean seed application to grow from zero now to 50%.

Investment Leading companies should be able to consolidate the fragmented agribusiness market and gain market share implications: with efficient products. We prefer Longping High-tech for its strong brand name and GM seed pipeline.

China's low per-person agricultural output. China's gross output per worker is just 1% of the US level, according to World Bank data. We think this is because China's farmland area per person is tiny. Also, farm efficiency is damaged by smallholding operations that have intensive labor requirements and are only semi-mechanized.

Exhibit 157: Exhibit 158: Agricultural gross output per worker in selective regions, 2017; China is China's arable land per person is tiny far behind

Source: World Bank, Morgan Stanley Research

102 M BLUEPAPER

Farmland consolidation a crucial reform. We believe China needs 3. Farmers' major source of income is not crops. Jobs in to consolidate this system of smallholdings into industrial-scale urban areas are now the major contributor to rural resi- farms given risks to food security, inferior farm economics, the aging dents' incomes, whereas in the 1980s over 80% of rural rural population, and ongoing urbanization. incomes came from farming. As salaries have continued to rise, many farmers have abandoned farming altogether. 1. Urbanization will depopulate rural areas. We believe 4. Increasing opportunity cost of unpaid labor. For the urbanization is unstoppable as the economy grows in any past decade, grain production needed an incentive policy to country. The urban population as a proportion of the total maintain good margins for farmers. The government's key population is still far below that of developed countries at incentive policy has been to raise grain prices to offset the more than 80%. increasing opportunity cost of unpaid labor (and fertilizer 2. Chinese farmers are migrant workers. Chinese farmers costs) and to keep the profitability of grain production do not focus solely on farming. Half of the rural population attractive. However, grain price hikes raise the burden on work in urban areas for most of the year and only return to the lower-income population, so we think this policy will their homes during festivals or busy farming seasons. not be sustained. Also, rising labor costs could be a long- term structural trend with the potential to impair China's food security.

Exhibit 159: Exhibit 160: Cash cost structure for rice production, 2017 Opportunity cost of unpaid labor is the biggest of all rice production costs

Source: NDRC, Morgan Stanley Research

MORGAN STANLEY RESEARCH 103 M BLUEPAPER Farmland consolidation to create efficiency. Because of its small- glyphosate-tolerant soybean was introduced to the market in 1996, er-scale farms, China's grain production has higher unit costs than US followed by GM corn in 1998. Afterwards, insect resistance was levels. China also has lower yields per hectare for corn and soybeans. developed as well as stacked traits – a combination of more than two If China opens up grain imports, domestic grain production would GM traits. sharply decrease. Soybeans are a good example of a crop where domestic production and planted area have been decreasing since As GM crops are able to reduce labor inputs, the new varieties have 2000. quickly surpassed conventional varieties over the past two decades, especially in the Americas. Currently, GM crops account for 81%, Genetically modified crops' penetration in the Americas. After 64%, 29%, and 23% of the global planted area of soybean, cotton, the commercialization of the herbicide glyphosate in the 1970s, corn, and canola. In the US – where GM crops originated – coverage Monsanto began developing glyphosate-tolerant crops through of GM crops is much higher, at 94%, 92% and 94% of planted areas genetic engineering with the intent of increasing glyphosate usage. A of soybean, corn and cotton.

Exhibit 161: Exhibit 162: China has higher unit costs but lower yields in corn production than the China has higher unit costs but lower yields in soybean production than US the US China Corn Production Cost 2014: yield=3,035kg/acre China Soybean Production Cost 2014: yield=872kg/acre As a % of As a % of Rmb/mu US$/acre Cent/kg total cost Rmb/mu US$/acre Cent/kg total cost Fertilizer 130 129 4.2 12% Fertilizer 47 46 5.3 7% Seed 55 54 1.8 5% Seed 39 38 4.4 6% Chemicals 15 15 0.5 1% Chemicals 16 16 1.8 2% Machinery leasing 105 104 3.4 10% Machinery leasing 77 76 8.7 11% Hired labor 28 28 0.9 3% Hired labor 20 19 2.2 3% Farmland rent 24 24 0.8 2% Farmland rent 65 64 7.3 10% Others 59 58 1.9 6% Others 25 25 2.8 4% Total cash cost 417 411 13.5 39% Total cash cost 288 284 32.5 43% Opportunity cost of unpaid labor 446 440 14.5 42% Opportunity cost of unpaid labor 197 194 22.3 30% Opportunity cost of land 200 198 6.5 19% Opportunity cost of land 183 180 20.7 27% All-in cost 1,064 1,049 34.6 100% All-in cost 667 658 75.4 100%

The US Corn Production Cost 2014: yield=4,320kg/acre The US Soybean Production Cost 2014: yield=1,307kg/acre As a % of As a % of Rmb/mu US$/acre Cent/kg total cost Rmb/mu US$/acre Cent/kg total cost Fertilizer 151 149 3.5 23% Fertilizer 38 37 2.9 8% Seed 102 101 2.3 15% Seed 61 60 4.6 13% Chemicals 30 29 0.7 4% Chemicals 28 27 2.1 6% Fuel, lube & power 33 33 0.8 5% Fuel, lube & power 22 22 1.7 5% Hired labor 3 3 0.1 0% Hired labor 3 3 0.2 1% Custom operation 19 18 0.4 3% Custom operation 10 10 0.8 2% Others 27 26 0.6 4% Others 24 23 1.8 5% Total cash cost 365 360 8.3 55% Total cash cost 186 183 14.0 41% Capital recovery of machinery 101 99 2.3 15% Capital recovery of machinery 89 87 6.7 20% Opportunity cost of unpaid labor 25 25 0.6 4% Opportunity cost of unpaid labor 18 18 1.4 4% Opportunity cost of land 178 176 4.1 27% Opportunity cost of land 161 158 12.1 35% All-in cost 669 660 15.3 100% All-in cost 454 447 34.2 100%

Difference % - China over the US Difference % - China over the US Cash cost 14% 14% 63% Cash cost 55% 55% 132% All-in cost 59% 59% 126% All-in cost 47% 47% 121%

Source: NDRC, USDA, Morgan Stanley Research Source: NDRC, USDA, Morgan Stanley Research

104 M BLUEPAPER GM crop application is just a matter of time in China. GM corn and Forecasting 50% GM seed penetration for corn and soybeans: soybeans have proved their safety as food. We think the current low China imports over 80% of its soybean demand and those imports adoption of GM crops is mainly because food supply independence are all genetically modified. Chinese consumers have been eating GM is a core part of China's food security policy. soybeans and corn for a long time. If urbanization and farmland consolidation continue, large-scale farming is needed to reduce labor We believe China's opposition to GM crops will end as the country inputs by more than half, and GM seeds are a must if China hopes to seeks to increase efficiency through industrial-scale farming. China's lower its farming costs and become competitive in the global market. conventional corn and soybeans will be largely replaced by GM vari- It is unrealistic to use labor to remove weeds on large farms, and in eties, in our opinion, as indicated by the acquisition of GM technolo- addition labor will be less available in future. We believe that China’s gies by Chinese companies in recent years. aggressive GM tech M&A in recent years shows it ultimately plans to adopt GM seed. Given that most of China's labor input on farms is Exhibit 163: mainly for weed removal and crop protection, we forecast a 50% GM GM crop penetration, by country penetration rate (GM seed sales as % of total seed annual sales) for corn and soybean by 2030 as China introduces large-scale farms.

Stock implications

Longping High-tech

We believe Longping High-tech is the most exposed to the agricultural modernization theme for the following reasons:

Source: USDA (United States Department of Agriculture), Morgan Stanley Research l Leading rice seed breeder. Longping is a diversified seed Exhibit 164: breeding company with a particular strength in rice. The com- GM corn, soybean and cotton planted areas (as a percentage of total pany's rice strain is the best in class in China and has the largest planted area) planted area among all hybrid strains, which has enabled the company to maintain a gross margin of around 45% over the past decade. l Benefits from farmland consolidation. Hybrid rice occupies half of the total rice planted area in China. Farmers who do not really focus on yield and resistant traits tend to prefer conventional strains, which can be collected and saved for the next planting. Nevertheless, we believe hybrid seeds will see accelerating penetration owing to industrial-scale farming. l GM pipeline is a significant opportunity. The Brazilian asset acquired in late 2017 greatly improved the company's GM technology and positioned it to benefit from China's coming GM

Source: USDA (United States Department of Agriculture), Morgan Stanley Research revolution.

MORGAN STANLEY RESEARCH 105 M BLUEPAPER Investment Theme #3: New Lifestyles in Smart Supercities

Exhibit 165: Investment Theme #3: Summary Key Beneficiary Key 2030 Forecasts Top Stocks

Railway construction companies focusing on Inter-city commuter rail: 17,000 km Transportation inter-city and metro rail build-up (vs. 2,000km now)

• CRRC Corp Ltd Annual housing price growth: 6% in five key city Developers with more landbank exposure to (1766.HK) Property clusters large cities and key cityclusters (vs. 4% elsewhere) • Haier Smart Home (600690.SS) Market share of top ten players in steel and cement: Materials Leaders in highly concentrated industries 60% and 70% (vs. 37% and 57% now) • Meituan Dianping (3690.HK) Companies with clear strategies for smart Smart home appliance sales: Consumer IoT home appliances and e-commerce leaders US$220bn (>5x vs. 2018) • New Oriental Education & Technology Group Top online tutoring and vocational education Vocational training market: US$300bn Education (EDU.N) players (3x 2018) • TAL Education Group Pharmas with strong, innovative pipelines, China's healthcare service market size: US$2.2trn (TAL.N) Healthcare and top healthcare service providers with (10.1% CAGR in 2015-30) solid growth potential • Jiangsu Hengrui Medicine (600276.SS) Gaming operators in Macau with bigger Gaming revenue: US$70-100bn Macau Gaming hotel capacity (>2x 2018) • Aier Eye Hospital (300015.SZ) Theme 3: New lifestyles in smart supercities Domestic annual tourism expenditure: US$1.5trn Tourism Top tourist destination operators (vs. US$0.78trn in 2018)

Source: Morgan Stanley Research

106 M BLUEPAPER 3a. Transportation

Overview: China already has the world's longest and fastest high-speed rail system, and it's not finished yet. We expect a further buildup of high-speed rail, inter-city rail, and metro lines over the next decade, making travel even more convenient for people in key city clusters.

Key forecasts: The high-speed rail system will grow from about 30,000km now to 65,000km by 2030, with most of the new construction focused on eastern and central China. Inter-city rail networks will increase from 2,000km to 17,000km, and the length of metro lines will reach 15,000km, up from about 5,800km in 2018.

Investment Within our coverage, CRRC and CRCC should benefit the most from Urbanization 2.0. CRRC is our top pick for its implications: exposure to rapidly growing inter-city and metro rail networks. CRCC is also exposed to this segment and has higher margins and returns than other China infrastructure plays.

China's high-speed rail mix is improving. Contrary to market per- Based on the HSR lines currently under construction and their com- ceptions, we believe HSR construction will focus on eastern pletion schedules, we project that 17,600km and 17,000km will be China rather than the mid-west over the next decade. completed in 2019-23 and 2024-30 in eastern, central, and western China, with respective distances of roughly 7,200km, 6,400km, and It is a widely held view that given the advanced HSR networks in the 4,000km in 2019-23 and 7,700km, 6,300km, and 3,000km in central and eastern parts of China, future HSR lines will be concen- 2024-30, accounting for 41%, 36%, and 23% of the total length in trated in western China, where the economy is less developed. The 2019-23 and 45%, 37% and 18% of the total length in 2024-30, vs. concern is that demand for HSR travel will be weaker in western 33%, 34% and 33% in 2014-18. This suggests that there will be an even China, resulting in lower train density allocated to those lines. larger proportion of HSR lines starting up eastern and central China. However, we regard this as a misunderstanding of the long-term outlook for the development of China's HSR network.

Exhibit 166: Breakdown of new operational HSR mileage in 2019-23, 2024-30 compared to 2014-18 2014~2018 2019e~2023e 2024e~2030e

18%, 23%, 3,047km 45% 33%, 33%, 4,000km 6,545km 6,499km 41%, 35%45%, 7,237km 7,663km 20% 37% 36% 6,290km 34% 6,440km 6,761km

Eastern Central Western Source: NDRC, Morgan Stanley Research estimates Eastern: Beijing, Tianjin, Hebei, Liaoning, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian, Guangdong, Guangxi, Hainan. Central: Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan. Western: Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, Sichuan, Chongqing, Yunnan, Guizhou, Tibet.

MORGAN STANLEY RESEARCH 107 M BLUEPAPER

Exhibit 167: Exhibit 168: New operational HSR mileage New operational HSR mileage, by region 7000 Km 100% 6% 19% 18% 16% 15% 18% 6000 90% 22% 80% 42% 45% 5000 48% 47% 2495 44% 27% 958 70% 4000 29% 40% 37% 910 764 60% 39% 632 53% 2025 3000 1952 50% 1677 1227 1727 432 28% 19% 2049 147 435 40% 2000 766 36% 1092 1090 1060 838 899 30% 46% 58% 50% 1000 2054 864 2077 1797 49% 45% 1758 1621 1435 20% 41% 42% 1040 1173 1211 1095 33% 471 30% 30% 0 138 10% 19% 2014 2015 2016 2017 2018 2019e 2020e 2021e 2022e 2023e 2024~30e 6% per year 0% Eastern Central Western 2014 2015 2016 2017 2018 2019e2020e2021e2022e2023e 2024~30e per year Source: NDRC, Morgan Stanley Research estimates Eastern Central Western Source: NDRC, Morgan Stanley Research estimates

Inter-city rail l Guangdong plans to construct 1,000km as part of its Pearl River Delta inter-city railway network. In addition to the '8-Vertical and 8-Horizontal' arterial HSR network, l In Jing-Jin-Ji, nine inter-city rail projects with a total length of China will build inter-city railway networks in several key provinces 1,100km will start construction by the end of 2020 in an and economic zones. Some target areas mentioned in the plan attempt to achieve a 0.5-1.0 hour commuting circle. include Beijing-Tianjin-Hebei, the Yangtze River Delta, and Pearl l In Chengdu-Chongqing, nine inter-city railway projects with a River Delta. total length of 1,000km are planned, of which 671km will be in Sichuan and 337km in Chongqing. Elsewhere in this report, our economics team highlights the five city l Hubei aims to construct a rapid inter-city passenger network clusters that have the most growth potential: the Guangdong-Hong with Wuhan at the center. According to the plan, 700km of Kong-Macao Greater Bay area, the Yangtze River Delta, Jing-Jin-Ji, the inter-city rail will go into operation in 2019-2030. Mid-Yangtze River area, and Chengdu-Chongqing. The team believes l Shandong has set inter-city rail targets of 700km by 2025 and it will be highly beneficial for economic development if there are 1,500km by 2034, up from 113km at the end of 2017. advanced inter-city railways to connect cities within the zones. Our confidence is also supported by the inter-city railway construc- We also highlight that some of the city clusters are very light on inter- tion plans announced by different provinces. Summing up all long- city rail. For example in Guangdong province, which contributed 11% term plans, we estimate that more than 15,000km of inter-city of China's total GDP in 2018, the total distance of HSR (including railway lines will be built in 2019-30. Based on the projections of new inter-city rail) was only 1,542km, or 5% of China's total, by the end of starts in the past and estimated completion schedules, we expect 2018. We believe these regions have significant potential to build 5,000km to be completed in 2019-23 and 10,000km will be com- inter-city networks in the future. pleted beyond 2023.

Some provinces within clusters have released aggressive inter-city In addition, the plan also mentioned some HSR extension lines that construction plans: will connect some smaller cities with inter-city or arterial HSR networks. We estimate that 6,570km of extension lines will be com- l Jiangsu expects to construct 980km of inter-city rail in the pleted in 2019-30. province. Including lines now under construction, the new completion length will reach over 2,200km in 2019-2030.

108 M BLUEPAPER

Exhibit 169: A glance at China's inter-city railways to be built in 2019-30, according to local governments' published plans 2,500 2,266

2,000

1,500 1,375 1,369 1,275 1,128 1,120 995 959 1,000 875 746 713 569 500 400 395 375 310 213 118 114 72 0

2019-2030 intercity railway new completion mileage (km)

Source: NDRC, Morgan Stanley Research

Some of the inter-city railway construction plans belong to 'long- decline in 2018 stemmed from NDRC's suspension of approvals for term plans' and construction may not be completed before 2030. new construction in August 2017 after a subway project in Baotou Nonetheless, we are optimistic on the outlook for inter-city rail, and was halted, with the approval process restarting in August 2018. we assume that in the period of 2024-30, 50% of current scheduled inter-city railways and other non-arterial HSRs will be completed. We believe increased urbanization must be accompanied by the rapid This suggests that over 2019-30, the annual average distance com- expansion of urban rail networks. In July 2018, the NDRC released a pleted will be 1,253km. Given that '8-Vertical and 8-Horizontal' will be document raising the economic hurdles cities need to clear before fully completed by 2030, we estimate total new HSR completions in undertaking subway and light rail construction. For subways, a city's 2019-30 will average 2,900km annually. public budget income and gross regional product (GRP) must exceed Rmb30bn and Rmb300bn, respectively, up from Rmb10bn and Buildup of metro rail lines to enhance inter-city and city-suburb Rmb100bn previously. For light rail, a city's public budget income and connectivity GRP must exceed Rmb15bn and Rmb150bn, up from Rmb6bn and Rmb60bn previously. We believe this will help China's urban rail net- Metro rail has enjoyed rapid development over the past decade works achieve healthy and sustainable development. owing to local government encouragement and the constant expansion of cities. In 2018, China's metro rail mileage (including subways, Stock implications light rail, monorail, rapid transit, trams, maglev transport, and APM) totalled 5,767km, covering total 35 cities, growing at a CAGR of 16.5% CRRC and CRCC since 2014. According to the China Urban Rail Transit Association, by the end of 2018 plans for urban rail networks in 63 cities had been approved, with construction already under way in 61 cities, with a Within our coverage, CRRC and CRCC should benefit most from total planned length of 7,611km. Urbanization 2.0. CRRC is our top pick for its exposure to rapidly growing inter-city and metro rail networks. CRCC is also exposed to We expect China's metro rail mileage in operation to grow by 18% in this segment and has higher margins and returns than other China 2019, 18% in 2020 and 16% in 2021, after a 16% dip in 2018. The infrastructure plays.

MORGAN STANLEY RESEARCH 109 M BLUEPAPER Exhibit 170: Exhibit 171: China's total metro rail mileage in operation to grow strongly in …while new mileage will maintain positive growth through 2023 2019-21... 1600 Km 52% 60% 1346 1371 50% 16000 Km 15000 25% 1400 38% 1276 40% 21% 28% 1207 14000 1200 30% 11977 19% 18% 18% 20% 1011 12000 1000 20% 16% 9260 10606 6% 869 16% 15% 16% 6% 6% 14% 15% 734 2% 10% 10000 15% 800 13% 0% 7984 573 8000 6777 600 -10% 421 448 430 5767 10% 6000 5021 400 -16% -20% 4152 3580 -39% -30% 4000 3132 200 5% -40% 2000 0 -50% 2014 2015 2016 2017 2018 2019e2020e2021e2022e2023e 2024~30e 0 0% per year 2014 2015 2016 2017 2018 2019e2020e2021e2022e2023e 2030e China Metro Rail Incremental Operating Length YoY

China Metro Rail Operating Length(Cumulative) YoY Source: China Urban Rail Transit Association, Morgan Stanley Research estimates Source: China Urban Rail Transit Association, Morgan Stanley Research estimates

Exhibit 172: China's metro rail mileage, by city China Metro Operating Length By Cities (Km) Cities Line number 2014 2015 2016 2017 2018 2019e 2020e 2021e 2022e 2023e Beijing 27 604 631 650 684 713 806 845 910 944 991 Tianjin 9 147 147 175 175 227 245 289 367 367 424 Shijiazhuang 2 0 0 28 28 41 54 70 78 78 Taiyuan 10 0 0 0 0 0 0 35 67 67 Hohhot 0 23 51 51 Dalian 9 127 167 167 181 181 205 249 262 282 282 Shenyang 11 114 121 125 125 130 166 198 198 198 213 Changchun 5 56 60 60 78 112 127 130 207 280 337 Harbin 5 17 17 17 22 22 30 56 56 88 123 Xi'an 6 52 52 89 89 124 130 154 219 239 259 Lanzhou 2 0 35 61 61 61 61 70 70 70 88 Urumqi 5 0 0 0 0 17 17 17 59 59 59 Qingdao 12 0 11 34 55 180 254 324 411 444 480 Jinan 4 0 0 0 0 0 0 26 62 81 117 Shanghai 27 613 653 683 731 784 784 821 860 900 938 Nanjing 14 187 232 232 365 365 365 381 418 466 506 Suzhou 9 70 70 86 138 165 209 209 253 287 356 Nantong 0 0 39 60 Huaian 1 20 20 20 20 20 20 20 20 20 Changzhou 4 000 0 34 54 54 54 54 Wuxi 6 56 56 56 56 56 61 90 114 133 133 Hangzhou 8 66 82 82 106 117 172 199 202 299 379 Taizhou 0 0 0 42 42 42 Ningbo 6 21 49 74 74 74 102 174 182 210 210 Shaoxing 0 0 24 24 59 Hefei 12 0 0 25 52 52 90 90 166 166 201 Wuhu 0 0 30 30 47 47 47 Fuzhou 7 0 0 9 25 25 25 53 57 98 122 Quanzhou 0 0 30 30 30 30 30 Xuzhou 5 000 0 22 46 64 64 64 Xiamen 5 0 0 0 30 30 72 109 124 147 261 Guangzhou 22 243 243 276 358 443 523 650 690 798 897 Foshan 7 21 27 34 34 34 66 66 66 132 175 Shenzhen 16 179 179 286 298 298 327 424 431 555 582 Dongguan 4 0 38 38 38 38 38 38 90 148 165 Zhuhai 1 999999 9 Nanning 6 0 0 32 53 84 84 110 137 159 197 Zhengzhou 6 25 69 89 134 134 174 200 230 289 319 Wuhan 12 96 123 179 251 351 372 436 468 516 576 Huangshi 0 0 19 19 38 38 38 Changsha 4 22 27 69 69 69 102 125 151 184 218 Luoyang 0 0 0 23 41 41 Nanchang 5 0 29 29 48 48 75 113 113 113 152 Chongqing 10 202 202 213 265 316 343 395 405 432 502 Chengdu 7 155 180 200 269 330 399 467 541 639 689 Kunming 6 59 60 63 86 86 86 154 171 189 222 Guiyang 4 0 0 0 13 35 63 63 91 91 147 Total 321 3132 3580 4152 5021 5767 6777 7984 9260 10606 11977

Source: China Urban Rail Transit Association, Morgan Stanley Research

110 M BLUEPAPER 3b. China Property

Overview: We believe property demand will be sustainable through 2030, as China's urbanization ratio rises from 60% to almost 75% by 2030, per our projection, on the back of smart cities and city clusters.

Key forecasts: We forecast incremental housing demand at 1,450mn sqm annually through 2030 (vs. 1,479mn sqm in 2018). Housing prices could increase by a 6% CAGR in city clusters and a 4% CAGR in non-city clusters.

Investment Companies with high landbank exposure to top-tier cities, such as Sunac, CR Land, and Longfor, should benefit implications: most.

Demographics suggest sustainable 3. Improvement in poor living conditions in lower-tier cities property demand through 2030 According to China's 2010 census, only 50% of households in tier 3-5 cities have tap water and showers. The lack of basic necessities in Property demand in China may have passed its 'golden age', but our lower-tier cities indicates the strong potential for upgrade demand in analysis of the country's demographics indicates demand should be China. relatively stable until 2030. 4. Key urbanization trends through 2030: 1. Shortage of urban housing stock We forecast incremental housing demand at 1,450mn sqm annually We estimate there were 244mn units of housing stock in 2018 for through 2030, driven by continuing urbanization, a shortage of 300mn urban households, indicating a shortage of 56mn units. urban housing, and improved living conditions in lower-tier cities. (Units of housing stock refers to houses with separate kitchens and bathrooms.) l Our economist expects the urban population to reach 875mn/970mn/1,056mn in 2020/25/30, representing urbaniza- 2. Continuous urbanization tion ratios of 62.3%/68.5%/74.5%, compared with 59.6% in 2018 China's urbanization rate was slightly under 60% in 2018, making it l We estimate the average size of urban households will drop similar to Japan's level in 1958. As Japan's urbanization rate was nearly from 2.77 people in 2018 to 2.6 in 2030, compared with 2.33 in 95% in 2018, we believe China still has significant potential for urban- Japan and 2.54 in the US currently ization. Our economist forecasts that China's urbanization rate will l We estimate the number of urban housing units per household increase to 74.5% in 2030, indicating an urban population increase of will increase from 0.81 in 2018 to 1.00 in 2030, which is still 19mn each year, providing solid support for urban property demand. lower than that in Japan (1.16) and the US (1.15) currently l Thus, we forecast residential demand at 1,450mn sqm per year through 2030, vs. 1,479mn sqm in 2018.

MORGAN STANLEY RESEARCH 111 M BLUEPAPER

Exhibit 173: Housing stock vs. number of urban households (mn) China urban residential housing stock China number of urban households (unit/household) China number of urban unit per household (RSH) 350 0.9

0.8 300 0.7 250 0.6

200 0.5

150 0.4

0.3 100 0.2 50 0.1

- 0.0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Source: CEIC, NBS, Morgan Stanley Research

China's urbanization process and the formation of metropolitan areas

China's urbanization ratio increased to 59.6% in 2018 but is still lower still lower than Japan's 72% in 2015. We believe economic growth and than that of developed countries. According to our economist's fore- the development of tertiary industries will be key drivers of further cast, China's urbanization rate will increase to 74.5% in 2030. We urbanization. believe the following factors are driving urbanization and population migration in China: 2. Income gap

1. Economic growth, especially in secondary and tertiary indus- We use the ratio of tier 1 cities' average wages to tier 5 cities' average tries wages as an indicator of the income gap in China ( Exhibit 177 ). Although the ratio decreased from 2.27x in 2005 to 1.82x in 2017, it China's GDP growth has slowed from 9.5% in 2011 to 6.6% in 2018 but is still relatively high, which could attract population inflows to large is still at a relatively high level. The contribution of tertiary industries cities. to total GDP increased from 40% in 2000 to 52% in 2018 but was

112 M BLUEPAPER

Exhibit 174: Exhibit 175: China's urbanization rate China's GDP growth vs. urban population growth (mn) China rural population (mn) Net increase of urban population China urban population 40 Real GDP growth (RHS) 20% 1,500 China urbanisation rate (RHS) 100% 30 15% 1,250 80%

1,000 20 10% 60% 750 10 5% 40% 500 0 0%

20% 250 -10 -5%

- 0% -20 -10% 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 2018 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 2018

Source: CEIC, Morgan Stanley Research Source: CEIC, Morgan Stanley Research

Exhibit 176: Exhibit 177: China's GDP breakdown by industry structure vs. urbanization rate China's income gap Tertiary GDP Secondary GDP (mn) Net increase of urban population Primary GDP Urbanisation rate (RHS) 100% 100% 40 T1 / T5 cities' average wages (RHS) 3.0 90% 90% 2.5 30 80% 80% 2.0 70% 70% 20 1.5 60% 60% 1.0 10 50% 50% 0.5 40% 40% 0 0.0 30% 30% -0.5 20% 20% -10 -1.0 10% 10% -20 -1.5 0% 0% 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015

1953 1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005 2009 2013 2017 Source: CEIC, Morgan Stanley Research Source: CEIC, Morgan Stanley Research

3. 1/5 cities have population inflows; large cities becoming larger l Large cities are becoming larger. From 2010 to 2017, tier 1 cities saw annual usual residents increases of 258k, vs. 103k in tier 2 Based on our analysis of China's 287 prefecture cities, we find the cities, 26k in tier 3 cities, and 12-14k in tier 4-5 cities. following key points: l The usual resident increase in higher-tier cities was mainly driven by migration instead of local urbanization. Usual resi- l Tier 1 & 2 cities have much larger usual residents than do low- dents over the registered population in 2017 was 76% in tier 1 er-tier cities. In 2017, tier 1 cities had an average population of cities and 19% in tier 2 cities in 2017, showing population 18.2mn, vs. 8.9mn in tier 2 cities, 4.9mn in tier 3 cities, 4.2mn in inflows. The ratio was 0%/-12%/-8% in tier 3/4/5 cities, indi- tier 4 cities, and 2.6mn in tier 5 cities. cating outflows.

MORGAN STANLEY RESEARCH 113 M BLUEPAPER

Exhibit 178: Exhibit 179: Usual residents in China's cities China's annual usual residents increase (mn people) 2000 2010 2017 (th people) 2000-10 2010-17 500 20 18.2 18 450 16 400 14 350 12 300 258 10 8.9 250 8 200 6 4.9 150 4.2 103 4 2.6 100 2 50 26 14 12 0 - Tier 1 Tier 2 Tier 3 Tier 4 Tier 5 Tier 1 Tier 2 Tier 3 Tier 4 Tier 5 Source: CEIC, China 2000 and 2010 Census, Morgan Stanley Research Source: CEIC, China 2000 and 2010 Census, Morgan Stanley Research

Exhibit 180: Exhibit 181: China's usual residents over registered population China's urbanization rate 90% 2000 2010 2017 2000 urbanisation rate 2010 urbanisation rate 76% 100% 80% 89% 90% 84% 70% 80% 60% 68% 50% 70% 60% 55% 40% 53% 50% 50% 30% 40% 19% 38% 38% 37% 20% 40% 28% 27% 10% 30% 0% 0% 20% -10% 10% -8% -20% -12% 0% Tier 1 Tier 2 Tier 3 Tier 4 Tier 5 Tier 1 Tier 2 Tier 3 Tier 4 Tier 5 National

Source: CEIC, China 2000 and 2010 Census, Morgan Stanley Research Source: CEIC, China 2000 and 2010 Census, Morgan Stanley Research

Exhibit 182: Exhibit 183: China's urbanization sources Percentage of China's new urban population that rent homes Urban population growth from local urbanisation, etc. Rent population net increase/ urban population net increase 60% 80% Urban population growth from migration 71% 70% 50% 3% 60% 40% 14% 50% 40% 29% 40% 30% 30% 36% 35% 20% 44% 20% 20% 34% 10% 10% 18% 0% 8% 8% -1% 0% 0% -10% Tier 1 Tier 2 Tier 3 Tier 4 Tier 5 Tier 1 Tier 2 Tier 3 Tier 4 Tier 5 Source: CEIC, China 2000 and 2010 Census, Morgan Stanley Research Source: CEIC, China 2000 and 2010 Census, Morgan Stanley Research

114 M BLUEPAPER Smart cities and city clusters lift Bay Area, such as Shenzhen, Guangzhou, Zhuhai, and Huizhou, have supportive urban redevelopment policies and allow developers to population capacity get land after redevelopment, which speeds up the progress of urban redevelopment. Smart cities City clusters The development of smart cities will help improve efficiency and the capacity to accommodate more people by alleviating issues caused City clusters are also an efficient way to improve population capacity, by growing populations, such as congestion and pollution. and the key is to develop high-speed rail and metro links to connect satellite cities with the core city. This requires infrastructure con- From a property perspective, smart cities can be built up by devel- struction and may dilute the attractiveness of central cities. oping ToD (Transit Adjacent Development) projects and underground cities, and developing more detailed and comprehensive city Regarding property price expectations, we expect prices in the five planning to enhance commuting efficiencies and increase population key city clusters to be more resilient given population inflows, strong densities. The key for the development of smart cities in central areas industry support and abundant resources (including healthcare and is to increase the plot ratio, as the current plot ratio in the central education). We expect 6% annual housing price CAGR in the five key areas of Beijing and Shanghai are relatively low, at around 2.5. city clusters, and 4% CAGR for other cities toward 2030. Increasing plot ratios involves redeveloping old buildings in central areas. However, urban redevelopment policies vary; in cities where Stock implications the local government is reliant on land sales revenue, the government will take back the redeveloped land and resell it in a public auc- tion at a much higher price. Under such circumstances, developers We prefer developers with high landbank exposure to higher-tier have less incentive to get involved in urban redevelopment projects, cities and city clusters due to: (1) solid population inflows and and, if governments have to do it on their own, it will take a longer industry support; (2) lower inventory risk; and (3) benefits from spill- time to redevelop the land. Currently, only some cities in the Greater over demand from nearby center cities and metro development.

Our top picks are Sunac, CR Land and Longfor.

Exhibit 184: China's smart cities 1. Smart cities in central areas: Development of ToD projects and underground cities could improve city efficiency and capacity to accommodate more people Enablers 2. City clusters: Development of high-speed rail and metro rail to connect satellite cities with hubs and build up city clusters 1. Increased plot ratios in central areas: The development of smart cities in central areas involves urban redevelopment, and in cities that are highly reliant on land sales revenue the government is often responsible for Hurdles urban redevelopment and progress can be slow 2. Dilution of central cities: The expansion of city clusters could dilute the attractiveness of central cities and may affect property prices there. This could discourage governments from promoting city cluster development. Beneficiaries Companies with high landbank exposure to tier 1 & 2 cities and strong execution: Sunac, CR Land, Longfor

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 115 M BLUEPAPER

Exhibit 185: Developers' landbanks, by city tier Overseas Tier 3-5 (others) Tier 3-5 (satellite) Tier 2 Tier 1 100% 4% 0% 3% 2% 9% 5% 8% 3% 8% 7% 13% 13% 11% 11% 90% 14% 16% 11% 1% 13% 7% 3% 10% 4% 13% 11% 8% 9% 8% 80% 7% 9% 34% 15% 7% 43% 27% 15% 15% 70% 29% 51% 73% 62% 43% 60% 54% 44% 11% 55% 50% 74% 61% 53% 21% 76% 54% 68% 69% 40% 58% 34% 69% 30% 60% 49%

20% 24% 39% 41% 32% 20% 34% 33% 10% 21% 25% 17% 15% 17% 10% 14% 11% 12% 9% 7% 7% 6% 0% CIFI Agile COLI Logan Sunac Vanke Poly A Poly CMSK Shimao Longfor Yuzhou GZ R&F GZ CR Land CR Gemdale Greenland SH Shimao SH Sino-Ocean Shenzhen Inv Shenzhen Country Garden Country Future Land A&H Land Future

Source: Company data, Morgan Stanley Research estimates

Exhibit 186: Developers' landbanks, by city cluster Central West Mid-Yangtze Greater Bay Area Yangtze River Delta Pan Bohai Rim 100% 6% 90% 7% 3% 80% 15% 0% 10% 10% 70% 5% 7% 14% 9% 10% 6% 60% 6% 4% 5% 7% 6% 13% 12% 55% 4% 25% 50% 20% 39% 18% 83% 20% 7% 8% 19% 2% 27% 5% 6% 40% 5% 78% 23% 43% 67% 6% 30% 23% 48% 15% 46% 16% 32% 40% 28% 30% 25% 20% 43% 25% 18% 20% 19% 8% 30% 13% 10% 20% 17% 12% 11% 13% 6% 7% 7% 10% 9% 9% 8% 9% 10% 0% 3% 4% 4% 0% 0% 0% CIFI Agile COLI Logan Sunac Vanke CMSK Poly A Poly Yuzhou Shimao Longfor GZ R&F GZ CR Land CR Gemdale Greenland SH Shimao SH Sino-Ocean Shenzhen Inv Shenzhen Country Garden Country Future Land A&H Land Future

Source: Company data, Morgan Stanley Research estimates

116 M BLUEPAPER 3c. Hong Kong Property Companies

Overview: Hong Kong developers have built mixed-used projects in key areas like West Kowloon and Island East, as well in tier 1 & 2 cities, such as Shanghai and Chengdu. They should continue to benefit as they build projects that enable higher population densities.

Investment SHKP and Hang Lung are the most exposed to quality mixed-used projects in tier 1 & 2 cities within China's key implications: city clusters.

What role do Hong Kong property development of new central business districts (CBDs) away from congested old city centers, speeding up integration. Moreover, well developers play in building city clusters? regarded hotels, better managed shopping areas and high-end resi- dences serve as necessary complements. Property companies in HK, History lessons from Hong Kong: Since the 1960s, HK has under- by leveraging their experience, can serve as developers and landlords gone several transformations, changing from an industrial hub to a (property management). global financial center. The city saw the formation of various new 'clusters' such as Tsim Sha Tsui in the 1960s, Central in the 1980s, the Policy support has been and will be in place: In the development reclaimed areas of HK Station and Kowloon Station in the 2000s, and of Qianhai, the government designated land sites for investment more recently HK Island East and New Kowloon East. We believe the property purposes to HK companies with experience in developing formation of city clusters in mainland China is in some ways similar and managing investment properties in Hong Kong of at least 1.5mn to the development of new economic clusters in HK, but on a much sqft GFA. This helped reduce competition in land auctions in favor of bigger scale. Commercial real estate is part of the infrastructure HK property companies. We expect additional supportive policy buildup. reforms, such as for foreign company holdings, capital repatriation and expat tax exemptions. Commercial real estate is the first building block: Apart from building homes, sustainable job prospects and economic develop- HK property companies have witnessed and participated in the devel- ment in these new areas are crucial to new migrants. Grade A offices opment of major CBDs in mainland China over the past 30 years, bring in multinational corporations and high-profile domestic com- including some of the most iconic buildings in city CBDs. Some exam- panies, which in turn attracts new migrants and brings back home- ples are: grown companies that had previously left. This helps facilitate the

MORGAN STANLEY RESEARCH 117 M BLUEPAPER Exhibit 187: Beneficiaries: Hong Kong companies with Highlighted developments of HK companies

Major CBDs/Iconic exposure to mainland China Time Related Company Developments China exposure: Among HK property companies, CK Asset has the Guomao area, Beijing 1978 Kerry Kuok Family/Shangri-la largest landbank in terms of GFA in mainland China. The company Shangri-la hotels in var- 1980s Kerry Kuok Family/Shangri-la had 96mn sqft of development property landbank as of December ious cities 2018, of which 40% is located in 'Chengdu Chongqing', one of the five Pudong IFC, Shanghai 2000s Sun Hung Kai key city clusters that we expect to continue attracting population

Sanlitun, Beijing 2000s Swire Prop inflows. CKA has 71% of its development landbank in tier 2 cities, and these areas should benefit as city clusters continue to expand. SHKP, Xujiahui, Shanghai 2000s Hang Lung Henderson and Sino also have over 50mn sqft of development land- Qianhai, Shenzhen 2010s Kerry, Wharf, New World bank in China.

Qiantan, Shanghai 2010s Shangri-la, Swire Prop Upside from China exposure: We estimate that there is potential Source: Company data, Morgan Stanley Research NAV upside of over 60% for Hong Kong real estate companies with Exhibit 188: China exposure, given that we expect property asset prices in tier 1 & HK property companies' landbanks in mainland China 2 cities (provincial capitals) to double by 2030. We identify Wharf,

GFA in sq.ft HLP, and Kerry as the most 'mainland China exposed' companies 96 100 with at least 50% of their assets in these cities. Wharf has invested 90 80 76 in the 'Chengdu Chongqing' cluster by building the high-profile IFS 70 66 60 complex with a luxury hotel, Grade A offices and discretionary malls. 49 50 40 Meanwhile, Hang Lung saw 1H19 retail sales in Wuxi (Yangzte River 40 32 31 29 30 24 Delta cluster) and Dalian grow 25-29% YoY, while malls in Tianjin 18 20 14 12 9 (Jing-Jin-Ji cluster) also saw a turnaround with positive sales growth 10 6 0 of 3% YoY. In Wuhan (Mid-Yangzte River Delta cluster) Spring City 66 HLP HLP CKA SHKP SHKP NWD NWD Kerry was opened in August 2019 and Wuhan and Hangzhou will be added Wharf Wharf HK HK Land Swire Prop Swire Henderson Henderson to the group's portfolio. Properties under development in China Existing investment properties in China

Source: Company data, Morgan Stanley Research *Data as of December 2018, Link REIT as of Sep-2018, no new breakdown disclosed

Exhibit 189: CKA's development landbank (96mn sqft as of December 2018)

Source: Company data, Morgan Stanley Research *Data as of December 2018, no new breakdown disclosed

118 M BLUEPAPER

Exhibit 190: Scenario analysis: Property asset prices in tier 1 and 2 cities to double by 2030

China Land bank in Total NAV China Exposure GFA HK$mn HK$mn % mn sq.m Rmb psm Wharf 164,810 87,748 53% 6.4 12,520 Kerry 105,190 45,581 43% 1.2 33,570 HLP 139,801 62,607 45% 3.9 14,665 HK Land (US$) 30,186 6,204 21% 6.2 7,189 Swire Prop 284,296 59,804 21% 0.9 61,020 SHKP 655,667 122,053 19% 5.9 18,946 CKA 363,196 50,738 14% 9.2 5,020 NWD 224,574 71,789 32% 9.9 6,609

Source: Company data, Morgan Stanley Research estimates

Exhibit 191: Exhibit 192: HK property companies' NAV exposure to China Wharf China's rental portfolio to continue to grow

China DP China IP HK property Others HK$bn China rental revenue China rental EBIT As % of NAV 5.0 100% 4.4 4.6 90% 4.5 4.1 80% 4.0 3.4 70% 3.5 60% 53% 45% 43% 3.0 2.6 2.5 2.6 50% 2.3 2.4 2.3 40% 32% 2.5 2.0 1.9 30% 21% 2.0 21% 19% 1.5 20% 14% 1.5 1.3 1.2 1.3 1.0 10% 1.0 0.8 0% 0.5 Wharf HLP Kerry NWD Swire HK Land SHKP CKA 0.0 Prop

Source: Company data, Morgan Stanley Research estimates Source: Company data, Morgan Stanley Research estimates

Exhibit 193: Exhibit 194: Hang Lung to add more retail malls in key city clusters Hang Lung China's rental portfolio to grow HK$bn Completion GFA (mn sq.ft) HLP China IP Opening Schedule China rental revenue China rental EBIT

45 7.0 5.8 40 38.4 6.0 36.2 Hangzhou Site 5.2 35 Wuhan Mall 5.0 4.5 31.2 5.0 4.2 4.2 3.9 4.0 4.0 30 Kunming Mall + 1 6.8 6.8 4.0 3.5 3.5 24.4 hotel + 2 Offices 3.2 25 2.8 2.7 2.7 2.8 3.0 2.7 2.5 2.5 20 Outside Shanghai 2.0 15 19.7 19.7 19.7 19.7 1.0 10 6.6 5 1.9 0.0 4.7 Shanghai 4.7 4.7 4.7 4.7 0 Mid-2010 Mid-2019 End-2019 2020 onwards 2024 onwards Est. Opening Source: Company data, Morgan Stanley Research estimates Source: Company data

MORGAN STANLEY RESEARCH 119 M BLUEPAPER

What could hinder growth? Hong Kong Exhibit 195: property developers tend to have a lower HK property companies have lower net gearings than their mainland peers appetite for high net-debt-to-equity ratios, and none of them has a gearing of over 50%. Their cautious mindsets could hinder their ability to aggressively bid for newly developed areas or any new satellite cities.

Source: Company data, Morgan Stanley Research. As of December 2018.

120 M BLUEPAPER 3d. Materials

Overview: We believe continued urbanization means demand from property and infrastructure will decline at a mild pace in the medium to long term, contrary to market expectations of a sharp decrease. In addition, industry consolidation should improve significantly by 2030.

Key forecasts: The steel industry should be much more consolidated by 2030, with the market share of the top 10 players reaching 60% from 37% now. We estimate the same ratio for the cement industry will hit 70% by 2030 from 57% now. Construction demand will be down around 6% in 2030 vs. 2019's level.

Investment Leaders in highly concentrated industries can offset the negative effects of a mild demand slowdown, including implications: Anhui Conch, CNBM and Baosteel.

Beneficiary of urbanization and improving The supply side should also see improvements. Since supply-side reform began in 2016, production capacity for many materials has infrastructure come under better control. However, many industries remain very fragmented. The government aims to push for industry consolidation Building materials: Our economics team expects China's urbaniza- as the next step in supply-side reform. Utilization in the steel industry tion ratio to rise from under 60% in 2018 to almost 75% by 2030. This has increased to 95% in 2019 from 60% in 2015, but pricing power is should sustain demand for urban housing. Our property team fore- still poor because of low concentration, at 37%. While the govern- casts incremental housing demand at 1,450mn sqm annually through ment aims to increase the top 10 players' market share to 60% by 2030, which is similar to the current level (1,479mn sqm for 2018). 2025, we think the actual progress might take longer to 2030 as over Our infrastructure team expect a further buildup of high-speed rail, 200mnt of capacity under different steel enterprises will need to be inter-city rail, and metro lines over the next decade, making travel consolidated to achieve the goal. For the cement segment, the even more convenient for people in key city clusters. market share of the top 10 players is currently 57%. Because leading players such as Anhui Conch are still planning to expand capacity and The market has been cautious on the long-term demand outlook for market share, and as China has banned building new capacity (enter- materials because of concerns about the property market and infra- prises can only do capacity swaps), we think their target can only be structure construction. In contrast, we expect demand will continue achieved by engaging in M&A, which would help raise the concentra- to be well supported over the medium to long term, with only a mild tion level to 70% by 2030. slowdown and not a sharp decline.

MORGAN STANLEY RESEARCH 121 M BLUEPAPER

Exhibit 196: Aluminum and copper to benefit from EVs and grid system CR10 of steel industry in 2018 upgrade: Our auto team expects EV penetration of 32.6% by 2030. This suggests more demand for aluminum (lightweight cars) and copper (more wiring in EVs, charging piles, and local grid and trans- former upgrades). More power grid investment should also translate into higher demand for copper, as power equipment accounts for around 40% of copper demand.

Contrary to market expectations of a sharp decline (30-50% slowdown) in demand, our analysis suggests the construction demand slowdown should be mild (a 6% decline in 2030 vs. 2019). First, we expect real GDP growth to slow regardless of the pace of transition in China’s growth model, which implies an accom-

Source: Company data, Morgan Stanley Research. panying deceleration in the old economy given its primary role in supporting growth in recent decades. Second, the change in growth mix – specifically as the investment share in the economy declines – also Exhibit 197: implies slower demand growth for sectors in the near term. CR10 of cement industry in 2018 Experiences from countries that have successfully made the economic transition also point to a more subdued demand outlook for CNBM, 22% the old economy in the longer term. For instance, trends in the steel industry across a number of developed markets suggest that per capita steel consumption tends to peak 10-20 years after the share Others, 43% of secondary industry in GDP has peaked. In China’s case, the ratio of secondary industry to GDP peaked around 2005, implying that China’s per capita steel consumption would have started to flatten Anhui Conch, 11% by now. From a per capita income perspective, although other countries’ historical consumption of various commodities shows that China’s per capita consumption has yet to peak, China’s age depen- Jidong Cement, 6% ACC, 1% CR Cement, 4% dency ratio has already inflected, contrary to the demographic Huaxin Cement, 3% Tianrui Cement, 2% Shanshui Cement, dynamics in these countries at around the same per capita income TCC, 2% Hongshi Cement, 3% 3% levels. The subsequent structural shift in the relative importance of Source: Company data, Digital Cement, Morgan Stanley Research investment and consumption will therefore imply more subdued commodity consumption.

Exhibit 198: Examples from developed countries indicate steel demand could decrease by 6-15% from 2019 to 2030 China's income per capita China's current income China's goal for 2030 13,535 16,187 24,704 in 2015 (US$) per capita (US$) (US$) China's current steel consum. per cap in 2030 consum. per cap at 555 522 474 consum. per capita (kg) at avg growth (kg) median growth (kg) Year for the country to Year for the country to Country's steel Country's steel reach income per capita of reach income per capita of consum. per capita in consum. per capita in Growth from YEAR US$ 16,187 - YEAR(1) US$ 24,704 - YEAR(2) YEAR(1) (kg/cap) YEAR(2) (kg/cap) (1) to YEAR (2) USA 1955 1972 621 638 3% S.Korea 1995 2005 562 726 29% Japan 1973 1988 588 479 -19% France 1968 1982 409 338 -17% UK 1968 1988 497 336 -32% Average 535 503 -6% Median 562 479 -15% Note: GDP per capita statistics are on purchasing power parity (PPP) and constant 2011 USD basis

Source: World Steel Association, NBS, Haver Analytics, Morgan Stanley Research estimates

122 M BLUEPAPER

Exhibit 199: Exhibit 200: In China, the share of secondary industry in GDP peaked in 2005 Sustainable steel consumption per capita: most developed economies range from 200-600kg/capita – China is within this range Steel consumption per cap (kg) Steel consumption per cap vs GDP per cap 1,400 China US Japan UK Germany France Australia Canada South Korea 1,200

1,000

800 South Korea

Japan 600 China Canada Germany US 400

France Australia 200 UK

- - 10,000 20,000 30,000 40,000 50,000

GDP per cap (US$, constant price at 2011 dollar)

Source: WSA, Japan Iron & Steel Federation, Korea Customs Service, World Bank, Morgan Stanley Source: World Steel Association, World Bank, Morgan Stanley Research Research. Note: Japan, South Korea and China steel consumption has been adjusted for steel content exported with auto, machinery and ships

Exhibit 201: Exhibit 202: China's cement consumption per capita is above international peers' China's coal consumption per capita leveling off at the higher end of sustainable levels global peers' sustainable levels Cement Coal consumption consumption per Cement consumption per cap vs GDP per cap per cap (kg of oil Coal consumption per cap vs GDP per cap cap (kg) equivalent) 2,000 China China 3,000 1,800 China US Japan US UK Germany France Japan Australia Canada South Korea Australia 1,600 UK 2,500 Germany South Korea 1,400 France Australia 2,000 1,200 Canada South Korea 1,000 US 1,500 Germany 800 Japan China South Korea 600 Australia 1,000 Germany Canada France 400 US UK 200 500 UK Canada Japan - US France - 10,000 20,000 30,000 40,000 50,000 - - 10,000 20,000 30,000 40,000 50,000 GDP per cap (US$, constant price at 2010 dollar) GDP per cap (US$, constant price at 2011 dollar) Source: Cembureau, World Bank, Morgan Stanley Research Source: BP Statistics, World Bank, Morgan Stanley Research

Exhibit 203: Exhibit 204: China's aluminum consumption per capita is still on an upward trajec- China's copper consumption per capita has more room to grow com- tory pared with global peers' sustainable levels Aluminum Refined copper consumption per cap Aluminum consumption per cap vs GDP per cap consumption per cap Refined copper consumption per cap vs GDP per cap (kg) (kg) 35 20 China US Japan UK South Korea 18 30 South Korea Germany France China Australia Canada Canada Germany 16 South Korea Australia 25 US 14 Germany Australia 20 12

10 15 Japan China Japan 8 France 10 France 6 Canada US UK 5 4 UK 2 - UK - 10,000 20,000 30,000 40,000 50,000 - - 10,000 20,000 30,000 40,000 50,000 60,000 GDP per cap (US$, constant price at 2011 dollar) GDP per cap (US$, constant price at 2011 dollar) Source: Woodmac, World Bank, Morgan Stanley Research Source: Woodmac, World Bank, Morgan Stanley Research

MORGAN STANLEY RESEARCH 123 M BLUEPAPER Exhibit 205: Morgan Stanley estimates of downstream demand drivers for different commodities: property and infrastructure are important drivers of demand Steel Cement Copper Aluminum Coal Property 42% 33% 9% 35% Infrastructure 25% 33% Auto 6% 10% 16% Power 47% 16% 53% White Goods 2% 15% 8% Electronics 8% 5% Machinery 15% Shipbuilding 3% Rural 33% Manufacturing Packaging 11% Metallurgy 18% Building Materials 8% Others 7% 11% 9% 21%

Source: Morgan Stanley Research estimates

Stock implications

Baosteel Chalco

Benefiting from smart city construction. Baosteel is one of the Continuous urbanization and demand for higher living stan- largest steelmakers in China, with a focus on high-end flat product. dards drives up aluminum consumption. Chalco is the largest alu- The company produces hot-rolled sheet, cold-rolled sheet, down- minum and alumina producer in China. Solid property and auto stream value-added coated products, seamless tubes and wire rods. demand through 2030 should help support aluminum consumption, In particular, Baosteel has the largest market share of auto sheet pro- as a respective 35% and 16% of demand is driven by these two seg- duction in China, which we think will benefit from the trend of shared ments. In addition, increasing white goods demand, along with mobility and electric vehicles, as mentioned by our autos team. increasing property demand, could further support aluminum Besides the construction of infrastructure, commercial real estate demand. Furthermore, China has capped the country's total alu- projects should also help create stable demand for steel in the minum capacity at 44-45mnt, which limits supply. medium term. Jiangxi Copper Anhui Conch Key beneficiary of copper price recovery driven by infrastruc- Well positioned for city cluster formation. Anhui Conch has ture stimulus. Jiangxi Copper is one of the largest copper smelters strong exposure to three of the four highly mentioned urban agglom- in China. China has stepped up its quantitative easing efforts heading erations in China: the Yangtze River Delta, the Guangdong-Hong into 2H19. The Ministry of Finance has required all special govern- Kong-Macau Greater Bay Area, and the Sichuan-Chongqing Urban ment bonds to be issued by end-September and all funds to be paid Agglomeration. The construction of infrastructure facilities in these to related projects by end-October. Part of the 2020 special bond regions, as well as property demand because of continual population quota will be front loaded to end-2019 to support infrastructure cap- inflows in these regions, will secure relatively stable cement ital expenditure. The infrastructure stimulus in China should signifi- demand. Because of limited new capacity announced in these cantly increase demand for copper, especially from the grid regions, we believe sustainable pricing will also ensure strong earn- investment side, which should benefit Jiangxi Copper. ings for Conch.

124 M BLUEPAPER 3e. Consumer IoT

Overview: In our view, urbanization will drive overall demand for home appliances. This will lead to sustainable growth in home appliances through 2030, with the penetration of smart appliances reaching 100% by then.

Key forecasts: We forecast a 2018-30 sales CAGR of 15% for IoT smart appliances given our outlook for 100% penetration in 2030 vs. 20% in 2018. This compares with a 2018-30 CAGR of 6% for the overall home appliances segment. In turn, the number of IoT appliances per household could rise to 7 units by 2030 (vs. just 1 today).

Investment Companies with clear strategies for IoT smart appliances should benefit from the trend, such as Haier Smart implications: Home, Haier Electronics, Midea, and Viomi.

Urbanization will provide a substantial growth opportunity for stages ( Exhibit 208 ). We are currently at Stage II, and the number China's consumer IoT segment, supported by advances in infrastruc- of IoT appliances per household is about 1 unit. By 2030, we expect ture, technology, and growing personal wealth. Currently, home to be somewhere in Stage IV, with the number of IoT appliances per appliances are the main use case for consumer IoT, with smartphones household reaching 7 units. and speakers serving as the interface to control them. Companies with clear IoT strategies should benefit from this trend. We expect China's home appliance market to reach annual sales of We believe key beneficiaries will include Haier Smart Home US$220bn by 2030, implying a 2018-30 CAGR of 6%. By 2030, all (600690.SS), Haier Electronics (1169.HK), Midea (000333.SZ), and home appliances will be 'smart', we believe, which implies a 15% Viomi (VIOT.N). CAGR. We expect smart home appliances to be adopted in four

Exhibit 206: Exhibit 207: Market size of China's home appliance industry Market size of China's IoT-enabled smart home appliances Rmb'bn Rmb'bn 1,800 1,602 1,800 1,602 1,600 1,600 1,400 1,400 1,200 996 1,200 906 1,000 810 818 819 1,000 800 616 660 800 615 529 600 600 421 315 400 400 201 117 200 200 71 - - 2015 2016 2017 2018 2019E 2020E 2021E 2030E 2015 2016 2017 2018 2019E 2020E 2021E 2030E White goods Small appliances Others White goods Small appliances Others

Source: iResearch, All View Cloud (AVC), Morgan Stanley Research estimates Source: iResearch, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 125 M BLUEPAPER Development of smart home appliances

Exhibit 208: Four stages of AI + IoT enabled smart appliances

Source: Morgan Stanley Research

Stage I: Isolated smart home appliances. Most of the appliances Stage III: Cross-scenario application – the genuine smart home. currently called 'smart' have achieved this stage. Historically, tradi- The next stage is to get all scenarios connected to achieve better tional appliances performed a singular function, and were not intelli- solutions for consumers. This requires better human-machine inter- gent or connected. In recent years, we have seen rapid growth in the action based on AI and IoT technologies. Daily life could be more use of connected home appliances that consumers control remotely easily supported by smart home appliances and high-speed data via mobile apps. Although these home devices are connected to the transmission. We would achieve real smart homes under Stage III. internet, they are generally isolated from one another, requiring consumers to download different mobile apps to operate them. Stage IV: Free connection of all products. Stage IV allows for the free connection of all equipment that a consumer uses in daily life, Stage II: Connected smart home appliances under a scenario. We including household electronics, home appliances and autos. The are currently at this stage of development, and we have seen many free connection is not confined to the home but can expand to the real applications of stages. A 'scenario' is a key concept in the smart living community, schools, offices, hospitals, etc. This is a stage of real home appliances industry. Locations of daily living scenarios at home connection across all scenarios within a person's daily life regardless include bedroom, living room, kitchen, bathroom, and dining room. of location. Due to infrastructure restrictions, we are not living in this From a solutions point of view, whole home solutions include air, stage. We expect a certain level of Stage IV to be achieved by 2030, water, laundry, security, health, and dining. Each scenario requires with the number of smart appliances per household reaching 7 units. many interconnected appliances. For example, in a living room scenario, door locks, lights, curtains, and air conditioners should be interconnected to achieve smart control and interact when people leave Participants in the smart home appliances or arrive home. Based on our estimates, the penetration of smart segment appliances is about 1 unit per household.

Smart home appliances are the main use case for consumer IoT at home. We place participants in the smart home appliances segment into four categories: tech hardware companies; internet companies (e-commerce and search engine); traditional home appliance companies; and traditional retailers.

126 M BLUEPAPER Strengths: Internet companies can provide the best support for con- Participants in the smart home appliances segment: nectivity and cloud computing, and they have the best customer l Tech hardware companies exposure, particularly for e-commerce companies, such as Amazon, ¡ China: Xiaomi, Huawei Alibaba, and JD, which control the sales channel. Tech companies gen- ¡ Global: Apple erally have strong product quality control and understand the needs l Internet companies – e-commerce of customers. ¡ China: Alibaba, JD.com ¡ Global: Amazon Weaknesses: Currently, internet companies are still burning cash to l Internet companies – search engine attract users and expand market share, but have been unable to find ¡ China: Baidu better ways to monetize the user traffic from IoT home appliances at ¡ Global: Google this stage. l Traditional home appliance companies ¡ China: Haier, Midea Traditional home appliance companies ¡ Global: Whirlpool, Electrolux, Samsung l Traditional retailers Traditional manufacturers have noted the importance of the growing ¡ China: Suning, GOME smart home appliance market. Companies with strong brand recognition and R&D capabilities have tapped into the market, targeting different customer and product mixes. In China, Haier is leading the Tech hardware and internet companies market as the first mover and can provide integrated services by connecting all devices it makes under Haier U+. Midea is also in the As internet traffic growth from computers and smartphones/pads is market using the M-Smart platform, but it focuses on small devices leveling off, internet companies are looking for new portals, such as such as safety items, cameras, lighting, and power switches. Gree has smart home products, to increase traffic from users. In the US, made less progress. In the global market, Samsung is the leader, with Amazon and Google are the leading players, with 90% of the smart its self-developed SmartThing platform. speaker market share, according to IDC. In addition, Apple is another large player in the market. In China, Alibaba, Xiaomi, Baidu, and JD are Traditional manufacturers usually provide closed ecosystems by key leading players, while we have seen limited progress from only connecting own-brand products at the early stage of IoT devel- Tencent. opment, particularly for large brands, such as Haier.

As tech and internet companies do not manufacture appliances, they Strengths: Appliance companies have better control over offline usually use self-branded smart speakers as the interface between distribution channels, which span the country. Distributors can help consumers and devices, and they integrate home appliances from with expansion, particularly in lower-tier cities. third-party devices. Tech and internet companies use their strong platforms in AI, big data, cloud computing, and e-commerce to estab- Weaknesses: Online sales channels are controlled by internet com- lish open ecosystems to attract home appliance brands and accel- panies (e-commerce) and distributors’ online platforms (Suning.com, erate consumer adoption. Entry barriers are high, and the space is GOME.com, etc). Home appliance hardware companies are at a disad- dominated by internet giants. vantage when competing with them, particularly internet companies. Hardware companies which have special focuses or strengths, such as Gree in air conditioning, may not be able to provide home integrated solutions.

MORGAN STANLEY RESEARCH 127 M BLUEPAPER

Exhibit 209: Summary of smart home appliance participants

Central con- Hardware Company Market position/strategy Platform Manufacturing troller focus Hardware companies Starting from Xiaomi smartphone, connecting prod- Xiaomi AI All home appli- Xiaomi Mi Home Own production + OEMs ucts from itself and ecosystem partners. speaker ances Huawei established HiLink Alliance by integrating Huawei router/ All home appli- Huawei top home appliance manufacturers. More than 50 HiLink No manufacturing app ances brands have joined HiLink Alliance. HomeKit lets people control connected accessories using Siri or the Home app on iPhone, iPad, and All home appli- Apple Watch. Many leading worldwide brands offer iPhone/iPad/ ances, with Apple accessories that are compatible with HomeKit and HomeKit AppleTV/Apple No manufacturing focus on small Apple devices. Watch devices Apple's HomeKit is currently available in China, and many appliance brands have joined the platform. Internet companies (AI/e-commerce) As the leading e-commerce platform and cloud No manufacturing (Alibaba Alibaba TmallGenie All home appli- Alibaba computing provider, Alibaba’s open source platform is an e-commerce and cloud speaker ances provides integrated solutions for home appliances. cloud computing provider) Baidu is an AI-oriented IoT platform provider, powered by the DuerOS system. Baidu has also signed All home appli- No manufacturing (Baidu is Baidu DuerOS Baidu speaker many strategic alliances with home appliance man- ances an AI solution provider) ufacturers such as Midea and Haier. All home appliances purchased from JDSmart No manufacturing (JD All home appli- JD (smarthome.jd.com) can be connected via JD JD Weilian NA Weilian is an e-commerce ances Weilian and platform provider) Focus on small Google uses the Google Home speaker as the voice devices like control center to connect third-party devices to No manufacturing (pro- Google Google Home safety products, Google Google’s platform. Google Home and related ser- viding portal for third-party Home speaker switches, lights, vices are not available in China. products) sensors, thermo- stats. Focus on small No manufacturing (pro- devices like Amazon uses Alexa, its cloud-based voice service, viding portal for third-party safety products, Amazon to connect devices via its Echo speaker. Alexa and Alexa Echo speaker products, and also enables switches, lights, related services are not available in China. e-commerce via own online sensors, thermo- channel) stats. Traditional home appliance companies Uses the U+ platform to connect Haier's own All home appli- Haier brands (including Haier, Casarte, GEA) and other U+ Inter-control Own production ances brands. Uses the M-Smart platform to connect Midea's Midea M-Smart M-Smart app Small devices Own production products. Connects Samsung and third-party products to the Own production + other Samsung SmartThing NA Small devices platform. brands Retailers Anything sold at Suning As sales channel providers, they integrate online/ NA NA No manufacturing the store offline sales platforms and provide an offline-experience/online-sales business model to consumers, Anything sold at GOME NA NA No manufacturing with a focus on one-stop shopping. the store

Source: Company data, Morgan Stanley Research

128 M BLUEPAPER Collaboration between different participants addition, consumers can also purchase Haier-branded smart products at Tmall (Alibaba) and JD.com, which are specifically designed for All participants are looking for new users/consumers to expand their the platform. businesses. However, the focuses of smart home appliance participants are different. Xiaomi, which started as a smartphone brand, is building a closed IoT system that is only compatible products made by Xiaomi and Xiaomi The purpose of internet companies is to build an open system to ecosystem companies. For example, Viomi, which is 33% owned by attract as many smart products and brands to the platform as pos- Xiaomi and Xiaomi's related party, started as one of Xiaomi's eco- sible. Traditional appliance companies have their own systems (such system partners, and launched its first product, a Xiaomi-branded as Haier's U+ and Midea's M-Smart) while also participating in other water purifier in 3Q15. Currently, Viomi provides Xiaomi-branded systems. For example, Haier, as the traditional home appliance manu- water purifiers, range hoods, and gas stoves to Xiaomi's IoT system. facturer, has its own smart platform Haier U+, which is a closed Xiaomi curates a wide range of additional products by investing in system for Haier's own products. However, Haier also collaborates and managing an ecosystem of over 200 companies, among which with other internet companies. The collaboration between Baidu's more than 100 companies are focused on the development of smart DuerOS and Haier U+ expanded the network for both companies. In hardware.

What does the participant have? What is the participant looking for?

l Strong technological capabilities (AI/IoT/5G) l Home appliances manufacturing capabilities l l Tech hardware companies Leading smartphone sales Brand position as a consumer IoT company l Large user base from sales of smartphones l Ecosystem, which could be a closed ecosystem

l Online platform and large user base l Internet companies (e-commerce and l Looking for new user traffic Ability to collaborate with different brands l search engines) Attracting more brands/products to the platform l Strong software capabilities

l Strong brand position as a home appliance l Product upgrading or premiumization l Home appliance companies company New growth driver as the traditional home appliances l Control over the whole production value chain sector is plateauing

MORGAN STANLEY RESEARCH 129 M BLUEPAPER Deep dive into 'scenarios' Exhibit 210: Smart kitchen scenario

'Scenarios' are a key concept of consumer IoT. Within each scenario, IoT companies are promoting bundled sales to consumers, meaning that consumers need to purchase all related products within the same platform/system to maximize the benefits. Scenarios at home can be defined under different categories: l By location: living room, kitchen, bathroom, bedroom, dining room. l By function: air treatment, water treatment, laundry, cleaning, Source: Morgan Stanley Research security, interaction, health. Exhibit 211: Exhibit 210 gives an example of a kitchen scenario at 6pm. The Interconnection of different scenarios smart refrigerator can suggest a recipe based on eating habits and food currently available in the refrigerator. The recipe can be trans- ferred to the screen above the stove to make cooking easier. The smart range hood can identify fumes and change power levels automatically. Meanwhile, the smart refrigerator can detect when food is running low and order online. When the food is ready, an alert can be issued through the screen above the stove or a smartphone.

The scenario in Exhibit 211 is achievable currently. In coming years, due to the variabilities of interaction and the wide application of smart appliances, smart functionality will not be confined to a single terminal or a single scenario. We expect interconnection among many scenarios. This should significantly improve the user experience when interacting with consumer IoT products. Source: Morgan Stanley Research

The ultimate purpose is to achieve the free connection of all consumer IoT equipment both inside and outside the home. In 2030, based on a unified connection platform and data protocol, smart homes, community services, and automobiles will be further connected and merged into a larger uni-scenario. The ultimate purpose is for consumers to have easier and better lives.

Entertainment Life management Life services Healthcare Home security monitor Smart washing machine Smart TV Smart locker Smart scale Home Smart refrigerator Smart speaker Smart door Smart blood pressure meter Smart air conditioner Home power management Car audio and entertainment Car security system Car air purifier system Automobile Car navigation system system Parking system Car alert system Car management Car services system Gate locker system Community Community activity management Convenience store retail system Healthcare services system Monitor system House cleaning services system Smart lighting system

130 BLUEPAPER

BottlenecksM in consumer IoT development Stock implications

Technological advances in AI, IoT and 5G have assisted the develop- Haier Smart Home (600690.SS) and Haier Electronics ment of consumer IoT, facilitating the implementation of different (1169.HK) scenarios and finally targeting a free-connection environment. However, we still see bottlenecks during the evolution of the con- Haier is a first mover and one of the leading smart appliance pro- sumer IoT industry. viders in China. In March 2014, Haier launched its U+ Smart Life Platform, in which smart home appliances interact with one another Compatibility among operating platforms: Connectivity is the key and with third-party services to provide 'smart life' solutions. The to IoT. Currently, different parties are running their own platforms launch of Haier's smart solutions initiative was ahead of similar such for smart appliances ( Exhibit 209 ), while appliances under dif- efforts from JD, Alibaba, Huawei, and Xiaomi, which were all launched ferent systems cannot connect to each other. Compatibility could be after February 2015. 'Going smart' encourages bundled sales of con- a bottleneck given that consumers may have their own brand prefer- nected smart home appliances to achieve better connectivity. In addi- ences when purchasing appliances. For example, Haier is the leader tion, Haier's 'smart' strategy is applied across all brands, which allows for washing machines and refrigerators, Gree is the top brand for air overseas consumers to purchase small home appliances under the conditioners, and Robam is the premium brand for kitchen appli- brands GEA, FPA, AQUA, and Candy. ances. A unified operating platform that can connect appliances from different platforms would accelerate the popularity of smart appli- Midea (000333.SZ) ances. Midea has been promoting its 'Smart Home + Smart Manufacturing' Key components/modules costs: Traditional home appliances have strategy. With continuing research and investments in AI, chips, sen- reduced the cost of consolidation in the production value chain. sors, big data, cloud computing, and other new technologies, Midea However, to implement smart functionality, the appliances should be has built the biggest AI team in the household appliances industry, installed with high-tech modules such as 5G modules, AI modules, which is committed to enabling products, machines, production pro- voice/image recognition modules. Currently, the cost of these mod- cesses, and systems that can sense, perceive, understand, and make ules is still high, making the price of smart appliances higher than tra- decisions in order to keep human-machine interactions to a minimum. ditional ones. Price reductions of key components/modules would raise the penetration of consumer IoT products. Viomi (VIOT.N)

Consumer switch costs: Large consumer IoT products such as home Viomi's main business consists of innovative IoT-enabled products appliances and automobiles have long lifecycles. It could take longer (including water purifiers, refrigerators, range hoods, gas stoves) for consumers to change from non-smart products to IoT products. together with a suite of complementary consumable products (e.g., In addition, home appliances and automobiles have a larger ticket water purifier filters) and value-added businesses (including both size than other consumer electronics. In the ultimate free-connec- hardware and services). Viomi, 33% owned by Xiaomi and Xiaomi's tion scenario, it is important that smart functionalities are applicable related party, is part of Xiaomi's ecosystem, initially focusing on in all scenarios, which requires solid infrastructure. water purifiers. Viomi currently provides Xiaomi-branded water purifiers, range hoods, and gas stoves to Xiaomi. As of June 30, 2019, Viomi had over 2.3mn household users. Exhibit 212: Smart home appliance statistics for Haier Smart Home, Midea, Xiaomi, and Viomi Devices Users Financials Revenue from smart home appliances: 2015: Rmb11.2bn Sold 28.3mn smart appliances in China from Haier Smart Home 60mn U+ Smart family users as of 2018 2016: Rmb19.6bn, +75% yoy, accounting for 15% of total revenue Jan'15 to Jun'18 2017: Rmb33.2bn, +69% yoy, accounting for 22% of total revenue 2018: Rmb59.7bn, +80% yoy, accounting for 33% of total revenue Revenue from ecosystem products of IoT segment: (1) 20.3mn Mi Home App MAU as of Dec'18 151mn connected devices (excl. smartphones 2016: Rmb9.0bn Xiaomi (2) 2.3mn users with more than 5 Xiaomi IoT devices and laptops) as of Dec'18 2017: Rmb15.0bn, +67% yoy, accounting for 13% of total revenue (excl. smartphones and laptops) as of Dec'18 2018: Rmb27.4bn, +83% yoy, accounting for 16% of total revenue Viomi total revenue: (1) 1.7mn household users as of Dec'18 Sold 2.5mn smart IoT products in China from 2016: Rmb313mn Viomi (2) 243,000 household users with more than 2 Jan'16 to Jun'18 2017: Rmb873mn, +179% yoy connected products oas of Dec'18 2018: Rmb2,561mn, +193 yoy

Source: Company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 131 M BLUEPAPER 3f. Education

Overview: Education demand is positively correlated with urbanization. We forecast a 7% revenue CAGR for the overall education industry in 2018-30, reaching US$1.9trn, and believe online and vocational education will be in strong demand owing to better technology and support for industry upgrades.

Key forecasts: We expect online K-12 tutoring revenues to grow fastest, rising 22x to US$160bn by 2030, thanks to more advanced technology application and an increase in penetration from less than 10% in 2018 to over 35% in 2030. (For details please refer to our online education report.) We estimate vocational education and training revenues will grow 3x, to US$300bn in 2030, driven by an increasing penetration rate.

Investment After-school tutoring will likely become more consolidated amid the city cluster trend; leaders EDU and TAL implications: should benefit most. TAL, as the largest online tutoring player, should also benefit from the increasing online tutoring trend. China Education Group is, in our view, best placed within our coverage to benefit from the vocational education trend.

We expect online education and vocational education will support Analysis of educational resources in the five key city and benefit from the growth of city clusters. clusters l Junior education: Access to quality educational resources is a Insufficient educational resources in the five leading city clus- key consideration in where families choose to work and live. ters, especially for primary schools and vocational education: With more online education and government policy support on Junior education is highly insufficient ( Exhibit 213 ) in some cities in promoting equalized junior education, we believe quality edu- the Greater Bay Area (such as Shenzhen, Dongguan and Foshan) and cation will be available to most city residents in 2030. the Yangtze Delta (such as Wuxi and Suzhou), even with the current l Higher education: As industry upgrades and urbanization create population. Overall, the average number of primary students per demand for more and better skilled workers, vocational educa- school in the five city clusters (786) is already 28% higher than the tion can directly connect skills training with job requirements national average level (612), and demand will only rise as populations to improve both knowledge and practical job training. The gov- in these clusters grow. ernment has emphasized supportive policies for vocational education in 2019 by enlarging enrollment quotas and providing We also expect stronger demand for vocational education, as urban- more fiscal support. ization and industry upgrades should increase the need for more and better skilled workers in these regions. According to the Ministry of Human Resources and Social Security, demand for highly skilled workers (level 8 technicians, the highest level among the current total of 165mn technical workers) in 2018 was double the supply of just 47.9mn. In 1H19, the percentage of high-tech manufacturing-added-value among total manufacturing-added-value was much higher in Guangdong (over 50%) and Beijing (over 40%) than the national average of 13.8%, highlighting a stronger need for skilled workers in these regions.

132 M BLUEPAPER

Exhibit 213: Exhibit 214: Insufficient primary and middle resources Greater Bay Area also has strong demand for higher education Average no. of K12 student per school (2017) Average no. of student per school (2017) 1,600 35,000 1,400 30,000 1,200 25,000 1,000 20,000 800 15,000 600 10,000 400 200 5,000 0 0 Jing-Jin-Ji Yangtze Greater Bay Mid-Yangtze Cheng-Yu Jing-Jin-Ji Yangtze Greater Bay Mid-Yangtze Cheng-Yu Area Area Primary School Middle School Higher education Higher vocational education National average - Primary National average - Middle National average - Higher education National average - Higher vocational

Source: CEIC, MoE, Morgan Stanley Research Source: CEIC, MoE, Morgan Stanley Research

Exhibit 215: Exhibit 216: Shortage of teachers in compulsory schools, and especially primary Developed areas like Beijing and Guangdong have higher % of high-tech schools manufacturing-added-value, implying higher demand for skilled Student to teacher ratio (2017) workers 25.4 % of High-tech mnufaturing-added-value among total manufaturing-added-value in 1H19 20.4 60% 53% 15.4 50% 40% 10.4 40%

5.4 30%

20% 0.4 14% Jing-Jin-Ji Yangtze Greater Bay Area Mid-Yangtze Cheng-Yu Primary School Middle School 10% National average - Primary National average - Middle 0% Source: CEIC, NBS, Morgan Stanley Research National Guangdong Beijing

Source: NBS, Provincial government.

Urbanization will boost after-school tutoring demand: As house- Exhibit 217: hold incomes and education levels are higher in major cities, parents Jing-Jin-Ji residents are the most likely to have a tertiary degree, and are more willing to pay for their children's education, driving up there is strong demand for AST in this region. We expect this to happen demand for K12 after-school tutoring (AST). In Beijing and Shanghai, in other city clusters, as well 45% and 32% of residents have at least a college degree – far higher % of population with college or above degree than the national average of 13% – and the cities' disposable incomes 30 19% 20% 18% per capita were Rmb62k and Rmb64k in 2018, compared with the 25 15% 16% 14% 20 13% 12% 12% national average of Rmb28k. Beijing and Shanghai have been big mar- 11% 15 10% kets for K12 AST over the past decade, despite having a lower average 8% 10 6% number of students per school. We believe this shows that higher 5 4% 2% income and education levels are supportive of strong K12 AST - 0% Jing-Jin-Ji Yangtze Greater Bay Mid-Yangtze Cheng-Yu demand. Area Population with college or above degree % of Bachelor among total

Given that we expect the five major city clusters to have a higher GDP Source: CEIC, NBS, Morgan Stanley Research per capita than the national average in 2030 – at Rmb181k and Rmb94k, respectively – as well as higher education levels as more educated workers move in, we anticipate stronger demand for K12 AST in these regions.

MORGAN STANLEY RESEARCH 133 M BLUEPAPER Population density improvements imply better learning center …empowered by new technology: The evolution of communica- coverage: For offline K12 AST, people usually go to learning centers tion infrastructure is accelerating the development of edtech and close to them (within 3-5km), and thus higher population densities in online education. Education companies are able to do in-depth data the city clusters imply that each learning center may be able to collection and analytics as more vendors offer technical support and accommodate more students. more cloud computing service providers offer on-demand IaaS, PaaS, and SaaS for educational products. Meanwhile, with higher penetra- Online learning will enable quality education resource tion and the faster iteration of smart devices, we expect more online sharing educational products with higher levels of interaction to be introduced to the market. The launch of 5G will also enable better learning Adaptive learning online could disrupt education... Learning experiences through live streaming, virtual reality, and other tech- should be personalized, but this is hard to achieve in offline settings nologies. because of limited resources, mainly teachers and physical locations. However, learning in offline groups is becoming more personalized and adaptive through online education. This is backed by the application of Artificial Intelligence in Education (AIED) in each component of the learning process. And the pace of change is accelerating, as seen in how business models are getting clearer after the past several years of trial and exploration.

Exhibit 218: IT infrastructure development accelerates application of edtech and online education products

Source: Morgan Stanley Research

Exhibit 219: Exhibit 220: spARk, an edtech start-up, uses AR technologies and tangible tool kits Squirrel AI Learning is devoted to improving study outcomes through to demonstrate wind power generation its adaptive learning system and has achieved very positive results

Competitors

Levels of difficulty for all 3-4 9 knowledge points

Number of tags for each 4-6 30+ knowledge point

Minimum questions in test bank 20mn 20,000 required for data analysis +26.18 points on +36.13 points on Score improvement* average average

*The test result is based on a teaching competition held in October 2017, when Squirrel AI compares teaching outcomes of a four-day middle school math bootcamp. The bootcamp is held for two groups of students who are taught by Squirrel AI's adaptive learning system and senior teachers who have more than 17 years teaching experiences.

Source: Company data, Morgan Stanley Research Source: Company data, Morgan Stanley Research

134 M BLUEPAPER Barriers to entry in online education will be significantly height- many edtech companies in the US, such as Knewton and DreamBox, ened… The online education market is developing ways to provide have chosen the B2B2C model. The data collection from students can high-level adaptive learning environments. Databases of, and algo- filter out more 'noise' stemming from non-study-related data, and it rithms for, students' learning behaviors will become core assets for is easier to track the same students' data over time. edtech companies, helping them widen their competitive moats over time. However, with the fast growth in online education user num- China has a large offline AST market. Thus, we expect the OMO bers – and therefore the fast accumulation of user data – we believe model to be successful in both the B2B2C and B2C scenarios. In the barriers to entry will rise higher and higher as the leaders advance in B2C scenario, we expect the leading market players, which have adaptive learning. broader offline learning center networks and high offline retention rates, can make the most of the OMO model. The large presence of ...and the OMO (Online Merges Offline) model will become more students in centralized classrooms can provide sufficient datasets, popular: As mentioned, the key to sustainable success in AI educa- and the companies bear lower customer acquisition costs in tion is to develop proprietary databases of, and algorithms for, launching new online education products. Meanwhile, smaller edu- learning behaviors. We think that cannot be achieved with pure cational companies and newcomers can seize opportunities with online education platforms. The main reason is that algorithms for B2B2C models to get access to target students. For instance, a learning behaviors have higher requirements for data collection in start-up featuring an AI tutor, Xizi-AI, cooperates with New Oriental terms of standardization, consistency, and granularity, and must be to gain massive amounts of analyzable data for faster product itera- rooted in observation of traditional offline education. That's why tion while saving on customer acquisition costs.

Exhibit 221: OMO model should enjoy high adoption rates in B2C and B2B2C scenarios B2C B2B2C C2B2C Boxfish, 17zuoye, Knowbox, B2B2C Offline LC only Small local AST institution TAL's Magic School, Xizi-AI, Online to Online Hujiang, GSX, Udemy (OMO available) Knewton, Dreambox

Online platform Tencent Classroom, Hujiang, LAIX, Xuebajun, Chegg, Byju B2B2C Online to Offline Qingqing only GSX, Coursera, Edx

Offline LC + Online platform EDU, TAL, Onesmart (OMO available)

Source: Morgan Stanley Research

MORGAN STANLEY RESEARCH 135 M BLUEPAPER Vocational education will be critical to supporting frontline workers by promoting the 'School-Company Cooperation' industry upgrades and urbanization initiative, which requires over 80% of large and medium companies to participate in vocational education through apprenticeship and Government emphasizing vocational education through subsi- internship programs by 2020. dies and preferential policies: Vocational education used to be seen as the place that the poorest performing students would end China is determined to promote skilled workers' social recogni- up, and was long neglected by society and government. However, tion and compensation: Learning from the experiences of Germany, vocational education will become more critical as urbanization accel- Japan, South Korea and Australia, the government is determined to erates tertiary industry development and as the modernization of promote skilled workers' educational attainment and social recogni- agriculture releases more of the rural workforce. tion by (1) establishing unimpeded pathways for vocational students to attain tertiary education and transfer to regular academic educa- In 2014, the government established preferential policies and tion ( Exhibit 223 ), and (2) establishing a national qualification binding targets to enhance the quality of vocational education. For system to include all vocational skill certificates and diplomas from example, the government set binding subsidy targets for both formal degree-granting institutions into a standardized and comparable and informal education, as shown in Exhibit 222 . It also emphasizes framework ( Exhibit 224 ). integration between vocational education and work experience for

Exhibit 222: Government's binding targets for formal and informal vocational education

Formal vocational education Informal vocational Secondary vocational On-site apprenticeship education Tertiary vocational education education programs

Additional 1mn new Enroll 500k students Enrollment All enrolled student All enrolled student 50mn enrolled students annually

Timeline By 2019 Since 2015 Since 2015 Since 2021 2019-2021

Reasonably higher than Min. Rmb12,000 per ≥Rb4,000 per studet Subsidy NA funding standard of Total Rmb100 bn student annually annually regular high schools

Source: MoE, Morgan Stanley Research

136 M BLUEPAPER

Exhibit 223: China intends to establish unimpeded pathways for vocational students to attain upper tertiary education and transfer to regular academic education

Source: MoE, Morgan Stanley Research

Exhibit 224: Guangdong has set up the first provincial qualifications framework to bridge credentials between formal and informal education Credentials in formal and informal education Qualifications Formal education Level Informal education Academic education Vocational education

Academic Doctors Vocational Doctors Level 7 - degree diploma degree diploma

Academic Masters Vocational Masters Level 6 1st Level degree diploma degree diplomas

Academic bachelor Vocational bachelor Level 5 2nd Level degree diploma degree diploma National Vocational Academic associate Vocational associate Qualification Certificates, Level 4 3rd Level degree diploma degree diploma Vocational trainings certificates Academic upper Vocational upper and others Level 3 secondary school secondary school 4th Level diploma diploma

Level 2 Lower secondary school diploma 5th Level

Level 1 Primary school diploma -

Source: Administration of Quality and Technology Supervision of Guangdong Province, Morgan Stanley Research

MORGAN STANLEY RESEARCH 137 M BLUEPAPER Location is key for vocational education: As the employment pros- Stock implications pects of graduates from vocational schools largely depend on local economic conditions, students will tend to choose good quality voca- TAL tional schools in developed regions. Vocational schools in China have been viewed as taking a backseat to bachelor's degree programs, and enrollment quotas at vocational schools have not been filled in We expect strong tutoring demand will be sustainable, supported by recent years in most provinces. Although the government has been rising incomes and higher tutoring participation rates. TAL is a leading encouraging the development of vocational education by targeting K12 AST institution, with both offline and online tutoring. With its an increase of 1mn in higher vocational student enrollment this year, well-established reputation and mature operations, we believe TAL we think the key beneficiaries will still be quality vocational schools will benefit, especially from its heavy investment in online tutoring, with high employment rates. which is a more scalable business and will likely be a more consolidated market. City clusters should be more attractive to students considering their higher GDP per capita and greater demand for skilled workers. The EDU State Council aims to establish 50 high-standard vocational schools by 2020 and encourages cooperation between vocational schools EDU is the largest tutoring institution in China and has the widest and enterprises, which is more likely to happen in developed regions. learning center network across different tier cities. It is especially Among the top five provinces that we have identified as the most competitive in secondary school student tutoring. We believe EDU is attractive for higher education, three are within the five city clusters best positioned to capture fast-growing demand from non-tier 1 (Guangdong, Jiangxi and Sichuan). cities, as well as demand from secondary students due to fiercer competition in the university entrance examination.

China Education Group

As a pure higher education operator with six of its nine schools located in the five key city clusters, we believe China Education Group will benefit from strong demand for higher education talent and skilled workers. It has the largest enrollment size among listed peers and the highest adjusted net margin, with a good track record of M&A execution.

138 M BLUEPAPER 3g. Healthcare

Overview: Urbanization will facilitate the formation of a tiered healthcare system in China by directing more patients to low-tier or private hospitals. The growing disposable income and an improving medical insurance system will favor the development of innovative drugs. The emergence of electronic prescriptions should also enable easier drug dispensing, especially in remote areas.

Key forecasts: We forecast the sizes of China's pharmaceutical and healthcare service markets to reach US$0.5trn and US$2.2trn by 2030, respectively, representing revenue CAGRs of 6.3% (2018-30) and 10% (2015-30).

Investment 1) Leaders in the pharmaceutical industry with a strong innovative pipeline, e.g., Jiangsu Hengrui Medicine implications: (600276.SS);

2) Healthcare service providers with leading market shares and growth potential in their specialties, e.g., Aier Eye Hospital (300015.SZ).

The rise of smart cities and clusters is aligned with the govern- format. The effort will require technologies in database management's plan to reform the healthcare system: One key pillar of ment, software programming, and cloud computing. healthcare reform is the tiered care system – redirecting patient flow from major urban centers back to local lower-tier hospitals. China's 2. Electronic prescription: The Chinese government has discussed 1,000 or so class 3 hospitals (the highest level) are overcrowded but allowing online drug prescription and filling but progress has been also have the best doctors and the latest equipment and technology. slow. If implemented, prescriptions are likely to be limited to Rural patients tend to travel to urban healthcare centers to seek common drugs that treat common chronic conditions. Critical condi- treatment. China currently lacks a referral system staffed by primary tions requiring more complex diagnostic algorithms will require face- care physicians (general practitioners) at the grassroots level. This to-face consultation with a doctor. The emergence of platforms like contrasts with more developed economies, where many common Ali Health and Ping An Good Doctor may enable online prescriptions conditions can be readily treated and managed at the community and drug dispensing. This will not only save trips to hospitals and level. As China provides more incentives for doctors to work at grass- clinics for patients, but enable hospital clusters to develop, espe- roots institutes, the patient burden on large urban hospitals can be cially for remote hospitals that may not necessarily have the alleviated as long as the enabling infrastructure is in place. resources to invest in hospital pharmacy space and dispensing systems. 1. Electronic patient records and filings: One key hurdle to over- come is transferring patient files (X-ray images, drug histories, prior 3. Telemedicine: This is closely related to patient record filing. Some diagnoses and surgeries) across hospitals. This avoids duplicate online platforms have been established, with patient forums, for testing efforts and frustration. The buildup of a national filing system patients to seek medical advice directly from doctors remotely. linking all public hospitals is as yet incomplete, not to mention Some platforms even allow video-conferencing with doctors for linkage to the county level, where the majority of patients live. A chronic, non-critical conditions. Telemedicine requires high band- number of large, university-affiliated hospitals have hospital clusters width connectivity, which may be lacking in remote areas due to around them, i.e., smaller hospitals that can refer patients to them, affordability. Remote imaging, storage, and computing, for example, whose systems are linked up with the parent hospital. These hubs are require supercomputers and real-time data transfer. The idea is to increasingly being developed with the help of private enterprise link up local imaging systems with imaging experts in large urban hos- management techniques. However, it remains a daunting task to keep pitals, so patients do not have to travel long distances to take tests. accurate medical histories of China's vast population in a digital This also requires a local certified technician who can operate the

MORGAN STANLEY RESEARCH 139 M BLUEPAPER equipment. Oftentimes, these are sales reps or distributor reps sent Stock implications by the manufacturer of the equipment locally. Other tests and vital signs can be taken with monitors or wearable devices. Jiangsu Hengrui Medicine

4. Wearables: Mobile patient monitors like glucose meters, oxygen meters and 24-hour Holters have been around for quite a long time. Jiangsu Hengrui Medicine is a pharmaceutical company focusing on Apps have also been developed for patients to download their time the development, manufacturing, and commercialization of series data onto USBs and transfer remotely to their physicians. oncology drugs, surgery drugs, and other pharmaceutical products. Connectivity not only leads to convenience but more accurate It holds a leading position in China in oncology and anesthesia drugs. tracking of vital signs. However, applications remain limited at the With a strong innovative oncology portfolio highlighted by anti-PD-1 moment. Diabetes (glucose level monitoring), cardiovascular (blood mAb camrelizumab and small molecule TKIs apatinib and pyrotinib, pressure monitoring), cardiology (episodes of syncope and brady- we think it will benefit significantly from the growth of disposable cardia) are well developed. But other critical conditions still require income in China, the improving medical insurance system that favors on-site testing and physical examinations. the development of innovative drugs, and the emergence of electronic prescriptions that should facilitate drug dispensing. The key goals of healthcare reform in China are to improve patient access to the best drugs available internationally, control reimburse- Aier Eye Hospital ment budgets, and provide basic healthcare to all. The large-scale urbanization in China over the past 50 years has focused medical Aier Eye Hospital is the largest private ophthalmology medical insti- resources on the large urban centers, where the patients are much tution in China. The company principally provides ophthalmology better insured through urban resident insurance and employers' medical services through the commercial mode of three-level insurance. The current reform aims to decentralize medical linkage of ophthalmocace diagnosis, ophthalmocace treatment, resources and distribute them more evenly. The development of pharmaceuticals distribution, and medical optometry. The company smart cities, with higher connectivity to smaller hospitals in hospital operated 300+ hospitals in China by the end of 2018. We think Aier clusters, can expedite this aspect of healthcare reform. is well positioned to exploit the growing demand from lower-tier markets, thanks to its continued efforts to establish a tiered network Key market size projections: We forecast China's pharmaceutical with ophthalmology hospitals, optometric clinics and eye care cen- market to grow at a 6.3% CAGR in 2018-30, from US$0.2trn in 2018 ters. to US$0.5trn in 2030, after 8.2% in 2014-18, in view of: 1) continued launches of innovative drugs vs. further cuts in generic drug prices by the government; 2) increasing affordability of innovative drugs thanks to growing disposable income and better insurance coverage; and 3) the emergence of electronic prescriptions, which should enable easier drug dispensing. We expect China's healthcare service market to grow at a 10.1% CAGR in 2015-30, to US$2.2trn by 2030, considering: 1) an 11.7% CAGR in China's aggregate hospital revenue in 2014-18, according to the NHFPC; 2) an emerging tiered healthcare system in China that could lead to the burgeoning of low-tier or private specialty hospitals; and 3) more diverse medical needs, such as medical aesthetics and assisted reproductive services.

140 M BLUEPAPER 3h. Macau Gaming

Overview: Macau's penetration rate in China is less than 2%, compared to 14% for Las Vegas in the US. With nominal GDP tripling by 2030, per our economics team's outlook, we expect Macau to benefit from improving infrastructure and rising affluence.

Key forecasts: We forecast that Macau's gaming revenue will double or more by 2030, from US$38bn to US$70-100bn. Furthermore, we believe Macau gaming's market cap could go as high as roughly US$300bn (3x the current market cap).

Investment Sands China, with the highest market share of hotel rooms, and Galaxy, with the biggest room supply increase implications: outlook, should benefit the most from China's visitation growth.

Growing in line with nominal GDP Mass as a percentage of revenue has gone up from 25% of the total in 2010 to 55% in 2018. Assuming mass contributes 75% of GGR and has a 3x higher margin, we estimate Macau's EBITDA at US$18-27bn Macau is the only place in greater China where gambling is allowed. in 2030, compared with US$10bn in 2018. Assuming a long-term There are six concessionaires that generated roughly US$38bn of average of 12x EV/EBITDA, this would imply a market cap range for gross gaming revenue (GGR) in 2018. GGR is strongly correlated with the six concessionaires of US$206-314bn compared with around China's nominal GDP (US$13.6tn in 2018) with an average ratio of US$100bn currently. Market cap upside would be even higher if we 0.34% since 2007. Based on our GDP projection range of assume higher EV/EBITDA multiples or lower FCFE yield as they start US$21.1-31.5tn by 2030, we estimate Macau's GGR will range from paying higher dividend yields and the risk of concession renewal goes US$70bn to US$100bn. This will be driven by growing GDP per away. capita and improved connectivity to Macau (high-speed rail, the HZMB bridge and the Hengqin development).

Exhibit 225: Exhibit 226: China's nominal GDP vs. Macau GGR China GDP per capita vs. Macau mass revenue per Chinese visitor 21% Growth YoY China nominal GDP (LHS) Growth YoY 150% 30% Growth YoY Growth YoY 40% Macau market cap (RHS) China GDP per capita (LHS) 19% 25% Mass revenue per Chinese visitor (RHS) 30% 100% 17% 20% 20% 15% 50% 13% 15% 10% 11% 0% 10% 0% 9% -50% 5% -10% 7%

5% -100% 0% -20% 2010 2011 2012 2013 2014 2015 2016 2017 2018 1Q11 3Q11 1Q12 3Q12 1Q13 3Q13 1Q14 3Q14 1Q15 3Q15 1Q16 3Q16 1Q17 3Q17 1Q18 3Q18 1Q19

Source: CEIC, DICJ, Morgan Stanley Research estimates Source:CEIC, DICJ, DSEC, Morgan Stanley Research estimates

MORGAN STANLEY RESEARCH 141 M BLUEPAPER

Exhibit 227: Macau's GGR, mass revenue and EBITDA are strongly correlated with China's nominal GDP

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Average China Nominal GDP US$ bn 3,550 4,597 5,110 6,102 7,568 8,565 9,682 10,449 10,962 11,201 12,140 13,616 Macau GGR US$ bn 10 14 15 24 33 38 45 44 29 28 33 38 Macau GGR as % to China Nominal GDP % 0.29% 0.30% 0.29% 0.39% 0.44% 0.44% 0.47% 0.42% 0.26% 0.25% 0.27% 0.28% 0.34% Macau Mass & slot revenue US$ bn 3 4 5 7 9 12 15 18 14 15 17 20 Macau Mass & slot as % to China Nominal GDP % 0.10% 0.10% 0.10% 0.11% 0.12% 0.14% 0.16% 0.17% 0.13% 0.13% 0.14% 0.15% 0.13% Macau EBITDA US$ bn 1.2 1.6 1.9 3.6 5.6 6.8 8.8 9.0 5.7 6.0 7.6 8.9 Macau Mass & slot as % to China Nominal GDP % 0.03% 0.04% 0.04% 0.06% 0.07% 0.08% 0.09% 0.09% 0.05% 0.05% 0.06% 0.07% 0.06%

Source: CEIC, DICJ, Morgan Stanley Research. Note: Average ratio of Macau EBITDA and China nominal GDP is from 2008, after the sector-wise EBITDA normalized.

Beneficiary of urbanization and improving started rising. Year to date, visitation from tier 1 cities (defined here as Shanghai, Beijing and Guangdong province) has grown by 22%, infrastructure higher than lower-tier city growth of 21%.

In our 2019 primer on Macau gaming, we highlighted that visitation Morgan Stanley Research now expects the high-speed rail network to Macau from lower-tier cities is rising. Between 2013 and 2018, visi- to expand more in eastern China (contrary to consensus expecta- tation from lower-tier cities has grown at a 7.5% CAGR, as compared tions), and thus we see more upside from tier 1 and new urban regions to 5% from tier 1 cities. However, since the opening of HZMB in (23 more cities to have more than 8mn population by 2030), driving October 2018, visitation from tier 1 cities, especially Guangdong, has visitation and revenue upside.

Exhibit 228: Exhibit 229: Visitation from higher-tier cities to start rising Visitation from tier 1 cities will start growing again 35.0 Total visitors in Macau (mn) 100% Visitor by tier of cities Visitors from lower tier cities (mn) 90% 30.0 Visitors from tier 1 cities (mn) 80% 51% Visitors from low-tier cities 70% 54% 53% 25.0 as a % of total Chinese 60% visitors 20.0 13.7 50% 15.0 40% 9.5 9.0 30% Visitors from tier 1 cities as 49% 46% 47% 10.0 7.4 20% a % of total Chinese visitors 11.6 10% 5.0 9.1 6.4 7.9 0% - 2013 2018 2019 YTD YTD YTD 2013 2014 2015 2016 2017 2018 2018 2019 Source: DSEC, Morgan Stanley Research. Note: tier 1 cities include Beijing, Shanghai and Guangdong Source: DSEC, Morgan Stanley Research. Note: tier 1 cities include Beijing, Shanghai and Guangdong province province

Exhibit 230: Tier 1 cities vs. lower-tier cities' visitation to Macau (Jan-Aug 2019) 2018 2019 2013 2014 2015 2016 2017 2018 YTD YTD Visitors from tier 1 cities (mn) 9.1 9.9 9.9 9.9 10.2 11.6 7.6 9.3 % as of Chinese visitors 49% 47% 48% 48% 46% 46% 46% 47% YoY growth 4% 9% -1% 0% 3% 14% 12% 22% Visitors from lower tier cities (mn) 9.5 11.3 10.5 10.6 12.0 13.7 8.8 10.3 % as of Chinese visitors 51% 53% 52% 52% 54% 54% 54% 53% YoY growth 17% 19% -7% 1% 13% 14% 16% 17%

Source: DSEC, Morgan Stanley Research. Note: tier 1 cities include Beijing, Shanghai and Guangdong province

142 M BLUEPAPER Stock implications ment will see the greatest upside. Sands China had 30% mass market share as of 2Q19 and is best positioned to capture the benefits of mass segment upside. We believe the key beneficiary of Urbanization 2.0 will be Sands China (the largest mass operator with the biggest room inventory), Hotel rooms allow guests to stay longer and potentially gamble followed by Galaxy (biggest planned increase in room supply). more. The number of hotel rooms can help operators gain mass market share. Sands China has been the market leader in terms of Mass revenue is largely a function of visitation and spending per hotel room supply (12.3k rooms in 2018), and we expect Galaxy to capita. With the improved connectivity and the emergence of city become the second-largest room supplier behind Sands by 2020, clusters, fueling GDP per capita growth, we believe the mass seg- with 5.2k rooms and 18% room market share.

Exhibit 231: Sands China is the largest mass operator with 30% mass market share Mass revenue Mass Mass revenue (US$ mn) 2014 2015 2016 2017 2018 CAGR 2014-18 2Q19 market share Sands China 4,910 3,816 3,867 4,646 5,405 2% 1,390 29% Wynn Macau 1,187 948 1,093 1,676 2,221 17% 580 12% MPEL 2,104 1,846 2,246 2,294 2,473 4% 723 15% SJM 3,830 2,849 2,616 2,656 2,978 -6% 801 17% Galaxy 2,122 2,068 2,489 2,908 3,320 12% 877 18% MGM China 1,229 975 992 1,050 1,407 3% 463 10% Macau 15,383 12,502 13,302 15,230 17,804 4% 4,834 100%

Source: Company data, Morgan Stanley Research

Exhibit 232: Sands and Galaxy have the biggest room inventories among the six operators Hotel rooms Hotel supply No of hotel room (average) 2014 2015 2016 2017 2018 CAGR 2014-18 2020e market share Sands China 9,296 9,296 10,598 12,696 12,360 7% 13,020 44% Wynn Macau 1,008 1,008 1,623 2,714 2,714 28% 2,714 9% MPEL 1,633 2,618 4,179 4,179 4,569 29% 3,987 13% SJM 821 821 821 821 821 0% 2,839 10% Galaxy 2,759 3,671 4,429 4,429 4,429 13% 5,220 18% MGM China 582 582 582 582 1,946 35% 2,009 7% Macau 16,099 17,996 22,232 25,421 26,839 14% 29,789 100%

Source: Company data, Morgan Stanley Research estimates

Exhibit 233: Hotel room growth driving mass revenue growth 60% Growth YoY Casino hotel room (ex satellite) 50% Mass and slot revenue 40% 30% 20% 10% 0% -10% -20% -30% 2Q12 4Q12 2Q13 4Q13 2Q14 4Q14 2Q15 4Q15 2Q16 4Q16 2Q17 4Q17 2Q18 4Q18 2Q19

Source: Company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 143 M BLUEPAPER 3i. Tourism

Overview: We believe urbanization will lift demand for sightseeing and holidays. Online travel agencies (OTAs), which are positioned as one-stop shops, are consumers' key booking channels for domestic travel.

Key forecasts: We estimate domestic tourism expenditure will grow at an 8% CAGR, from US$0.78trn in 2018 to US$0.9trn in 2020, and then at a 5% CAGR, to US$1.5trn to 2030. We expect domestic travellers to grow at a 10% CAGR, from 5.5trn trips in 2018 to 6.7trn trips in 2020, and then at a 4% CAGR, to 10trn trips to 2030. Online penetration of travel booking (hotels, transportation and others travel related products) grew from 11% in 2013 to 37% in 2018, and we expect that to reach 46% by 2022.

Investment Tourist destination operators like CYTS and OTAs such as Ctrip and Tongcheng-Elong should benefit. implications:

Domestic tourism expenditure to almost double to US$1.5trn by Exhibit 234: 2030: Our expectation was based on: China's domestic travel spending to reach US$1.5trn by 2030

Domestic Travel Expenditures (US$bn) l Real GDP per capita: Domestic travel frequency correlates 1,800 strongly with real GDP per capita. We expect travel frequency 1,600 MS Est. 2030(E), 1,540 to improve from 4x in 2018 to 7x by 2030, a level similar to 1,400 1,200 that of US during 2015-17. 1,000 l Travel expenditure as % of GDP: We expect this to gradually 800 2020(E), 905 decrease from 1.4% in 2018 to 0.8% by 2030, similar to that of 600 2018, 777 US during 2015-17. Higher travel frequency could means shorter 400 200 stay per travel or lesser spending each time. Chinas domestic - travel expenditure as a % of GDP as been declining gradually * from 4.6% in 2002 to 2.0% in 2001 to 1.4% in 2018.

Source: China Ministry of Culture and Tourism, Morgan Stanley Research estimates

144 M BLUEPAPER Growth in short leisure trips

The development of city clusters and improving transportation networks should help promote short trips to tourist destinations in city clusters. We believe growing household incomes and the rise of the middle class will increase the frequency of domestic travel ( Exhibit 235 ). In addition, China's domestic travel frequency of 4x per year in 2018 was lower than the US level of 7x. We expect China's domestic travel frequency to catch up to the US level by 2030 ( Exhibit 236 ).

Exhibit 235: Exhibit 236: Domestic travel frequency is highly correlated with GDP Domestic travel frequency per person – China vs. the US

Per Person Domestic Travel Trips vs Real GDP in China (1995- Times/Year Domestic Travel Trips Per Person Per Annum 4.5 2018) Trips / Persons 8.0 4.0 2018, 7.0 2 7.0 3.5 y = 8E-06x + 0.0005x + 0.3608 R² = 0.9986 6.0 3.0 5.0 China 2018, 4.0 2.5 4.0 US 2.0 3.0

1.5 2.0 MS Est. 1.0 1.0 0.5 - - - 100 200 300 400 500 600 700 Real GDP Index (1995=100) Source: China Ministry of Culture and Tourism, US Travel Association, Morgan Stanley Research

Source: China Ministry of Culture and Tourism, US Travel Association, China National Bureau of Statistics, US Bureau of Economic Analysis, Morgan Stanley Research

Urbanization promotes sightseeing

Urban residents tend to travel more for sightseeing and holiday purposes than people in rural areas, according to a survey by the China Tourism Academy in 2017 ( Exhibit 237 ).

Exhibit 237: Domestic travel by purpose of visit, 2017

Domestic Travel % By Purpose 60% Urban Citizens Rural Citizens 48% 50% 40% 36% 32% 30% 27%

20% 14% 13% 8% 8% 10% 6% 5% 1% 1% 0% Others Holiday Medical Business Family visit Family Sightseeing

Source: China Tourism Academy, Morgan Stanley Research

MORGAN STANLEY RESEARCH 145 M BLUEPAPER Stock implications Exhibit 238: Online penetration of overall travel booking market will continue to rise, benefiting the OTAs CYTS 50.0% 45.7% Online penetration of 45.0% 45.0% 43.4% China travel booking 40.9% 40.0% market 36.9% To benefit from Jing-Jin-Ji and Yangtze River Delta cluster devel- 35.0% 31.5% opment: Over 95% of CYTS's profits from 1H16 to 1H19 came from 30.0% 25.8% two tourist destinations – Wuzhen in the Yangtze River Delta (near 25.0% 19.7% Shanghai and Hangzhou) and Gubei in Jing-Jin-Ji (near Beijing). CYTS 20.0% 13.7% 15.0% is developing a third destination, Puyuan, near Wuzhen. The com- 10.6% 10.0% pany should benefit from short-term leisure travel growth as city 5.0% clusters develop with better infrastructure. In particular, we believe 0.0% that Gubei has strong growth potential. Gubei has the same area as 2013 2014 2015 2016 2017 2018 2019E 2020E 2021E 2022E Source: iResearch, Morgan Stanley Research estimates Wuzhen, but in 1H19 its visitation level was only 23% of Wuzhen's, and its net profit was only 14% of Wuzhen's. Exhibit 239: Ctrip and Tongcheng-Elong Online penetration of travel booking by segment 100% Air ticket Travel market GMV online penetration 89% 90% Train ticket 80% Ctrip is the leading OTA in China and offers the most comprehensive 80% 75% Bus ticket 71% travel products and round-the-clock customer service including 70% 67% Hotels 61% more than 60 travel-related value added services (VAS) in and out- 60% side of China. Tongcheng-Elong, which has about one-third the 50% 40% 31.6% domestic room night market share of Ctrip and offers over 30 travel- 26.7% 30% 21.5% related VAS, is geared toward domestic travel and lower-tier cities. 20% We believe rising urbanization in China and new transportation infra- 10% 1% 1% 2% structure will drive inorganic growth for the OTAs as 1) online pene- 0% 2015 2016 2017 tration of travel booking has risen from 11% in 2013 to 37% in 2018 Source: iResearch, Morgan Stanley Research and will hit 46% in 2022, we estimate, 2) OTAs remain the key channel for travel booking, according to our AlphaWise survey in Exhibit 240: May, and 3) OTAs are positioned as one-stop shops, providing travel OTAs remain the key channel for consumers to book travel products from 'pre-departure' to 'on the road' to 'arrival'.

Source: AlphaWise, Morgan Stanley Research estimates

146 M BLUEPAPER

Exhibit 241: One-stop shop services provided by OTAs

Pre-departure On the road Arrival

express ticket booking meals accommodation Monitor ticket availability and automatically pur- Order ahead and have meals delivered on board Large and diversified offerings catering to users' chase tickets at specified time slots and price directly budgets and preferences ranges Ticket delivery Lounge Attraction ticketing Deliver tickets to doorstep by counter Access to lounges at airports and train stations Book value-for-money ticket packages online Reservation transfer Pickup Car hire Transfer accommodation reservations to other Airport/train station pickup service Online taxi/car booking users Travel solution Social Social Cross-sell accommodation and transportation Networking with people you meet during the Share review and personal travel experience products journey online

Source: Company data, Morgan Stanley Research

MORGAN STANLEY RESEARCH 147 M BLUEPAPER Summary of Stocks Exposed to Urbanization 2.0

Exhibit 242: Summary* of stocks exposed to Urbanization 2.0 (ranked by industry group, then market cap)

Avg daily Return on equity Ticker Company name Analyst (Primary) Ind Group Price target Share price Mkt cap. Price to earnings Price to book EPS Growth (%) Dividend yield (%) t/o (ROE) (%)

Local ccy US$ mn US$ mn 2019e 2020e 2019e 2020e 2019e 2020e 2019e 2020e 2019e 2020e 600104.SS SAIC Motor Corp. Ltd. Yeung, Jack Auto & Comp 30.0 23.8 104 38,870 7.7x 7.4x 1.1x 1.0x 15.4% 14.7% (0.1%) 4.4% 5.3% 5.5% 2238.HK Guangzhou Automobile Group Yeung, Jack Auto & Comp 10.5 7.6 30 15,208 7.3x 6.5x 0.9x 0.8x 12.7% 13.2% (11.1%) 12.3% 5.5% 4.1% 600741.SS Huayu Automotive Hsiao, Tim Auto & Comp 24.0 23.5 39 10,366 10.4x 8.9x 1.6x 1.4x 15.7% 18.2% (11.3%) 16.5% 4.3% 5.0% 0489.HK Dongfeng Motor Group Yeung, Jack Auto & Comp 10.0 7.4 12 8,089 4.0x 3.8x 0.4x 0.4x 12.2% 11.7% 11.7% 5.5% 3.2% 3.6% 1316.HK Nexteer Automotive Group Hsiao, Tim Auto & Comp 8.0 6.2 5 1,991 7.6x 7.0x 1.2x 1.1x 15.7% 16.6% (31.0%) 8.7% 2.7% 2.9% 3968.HK China Merchants Bank Xu, Richard Banks 46.4 37.4 93 122,165 9.3x 8.1x 1.4x 1.3x 17.1% 17.4% 14.6% 14.5% 3.2% 3.7% 600036.SS China Merchants Bank Xu, Richard Banks 40.7 34.8 238 122,165 9.5x 8.3x 1.4x 1.3x 17.1% 17.4% 14.6% 14.5% 3.2% 3.6% 601166.SS Industrial Bank Co. Ltd. Xu, Richard Banks 26.8 17.5 173 46,727 5.5x 4.8x 0.8x 0.7x 15.0% 15.6% 8.7% 14.8% 4.3% 4.9% 000001.SZ Ping An Bank Xu, Richard Banks 18.8 15.6 184 37,450 9.2x 7.7x 1.1x 0.9x 13.3% 13.9% 17.6% 18.5% 1.1% 1.3% 1766.HK CRRC Corp Ltd Luo, Kevin CapGoods 10.4 5.3 16 26,436 10.2x 8.7x 1.0x 0.9x 10.6% 11.5% 20.9% 17.4% 3.7% 4.4% 1186.HK China Railway Construction Luo, Kevin CapGoods 11.6 8.7 16 17,537 5.2x 4.7x 0.6x 0.5x 12.1% 12.3% 14.8% 12.5% 2.9% 3.2% 601186.SS China Railway Construction Luo, Kevin CapGoods 12.9 9.5 114 17,537 6.2x 5.5x 0.7x 0.6x 12.1% 12.3% 14.8% 12.5% 2.4% 2.7% 600406.SSNARI Technology Hou, Eva CapGoods 26.7 20.5 47 13,224 19.2x 15.9x 3.0x 2.7x 17.8% 19.0% 17.5% 20.5% 1.8% 2.1% 2208.HK Goldwind Hou, Eva CapGoods 11.7 9.5 7 6,983 12.3x 11.2x 1.2x 1.1x 11.3% 10.3% (22.4%) 10.1% 2.2% 2.5% 002202.SZ Goldwind Hou, Eva CapGoods 14.7 12.5 50 6,983 18.0x 16.2x 1.7x 1.6x 11.3% 10.3% (23.2%) 11.3% 1.5% 1.7% 0552.HK China Communication Service Co Ltd Liu, Yang CapGoods 7.0 4.5 11 3,958 9.1x 8.4x 0.8x 0.8x 9.6% 9.8% 7.1% 8.4% 4.0% 4.3% 000333.SZ Midea Group Co Ltd. Lou, Lillian Cons Durables 49.6 51.1 208 47,635 15.1x 13.1x 3.5x 3.0x 27.2% 26.8% 11.3% 15.2% 2.5% 3.1% 600690.SS Haier Smart Home Co Ltd Lou, Lillian Cons Durables 20.8 15.3 71 13,632 11.7x 10.0x 2.1x 1.8x 21.2% 21.5% 10.8% 17.1% 2.6% 3.0% 1169.HK Haier Electronics Group Co Ltd Fan, Hanli Cons Durables 25.0 21.2 14 7,580 12.7x 10.9x 2.0x 1.7x 17.9% 18.3% 11.6% 16.2% 2.0% 2.3% 002405.SZ NavInfo Co Ltd Hsiao, Tim Cons Durables 12.5 16.3 106 4,474 71.2x 53.9x 4.3x 4.0x 6.2% 7.9% (6.2%) 32.0% 0.5% 0.7% VIOT.O Viomi Technology Co Ltd Lou, Lillian Cons Durables 12.9 7.9 1 547 10.1x 6.0x 2.7x 1.9x 36.1% 44.5% 82.9% 67.4% 0.0% 0.0% 1928.HK Sands China Ltd. Choudhary, Praveen Consumer Svs 40.0 35.4 61 36,594 16.6x 14.3x 8.1x 7.3x 50.0% 56.3% 6.6% 16.0% 5.7% 5.7% 0027.HK Galaxy Entertainment Choudhary, Praveen Consumer Svs 52.0 48.7 83 26,979 15.2x 14.5x 2.9x 2.6x 22.2% 20.1% (1.9%) 4.8% 2.0% 2.1% TAL.N TAL Education Group Zhong, Sheng Consumer Svs 43.0 37.7 30 22,651 57.7x 47.5x 9.1x 8.0x 22.9% 19.4% 56.2% 28.7% 0.0% 0.0% EDU.N New Oriental Education &Technology Group Zhong, Sheng Consumer Svs 120.0 114.3 29 18,170 32.8x 33.3x 5.8x 6.3x 20.8% 23.2% 16.9% 31.4% NA NA 0839.HK China Education Group Holdings Ltd Zhong, Sheng Consumer Svs 14.9 11.8 7 3,035 28.1x 22.6x 3.1x 2.8x 12.4% 14.0% 57.0% 26.4% NA NA 600138.SS China CYTS Tours Holding Co Ltd Ling, Hildy Consumer Svs 16.0 12.1 21 1,224 13.3x 12.1x 1.3x 1.2x 10.8% 10.8% 9.2% 9.7% 1.3% 1.4% 000998.SZ Yuan Longping High-tech Agricultural Lu, Jack Food, Bev. & Tob 18.1 12.5 36 2,204 16.5x 14.3x 2.2x 2.1x 14.7% 15.7% 19.7% 15.8% 3.2% 3.7% 300015.SZ Aier Eye Hospital Group Hu, Yolanda HthCare 37.0 35.5 56 15,451 65.9x 53.5x 16.6x 13.9x 29.2% 30.7% 39.7% 23.3% 0.4% 0.5% 2318.HK Ping An Insurance Company Jiang, Jenny Insurance 111.0 90.1 362 211,424 9.6x 9.6x 2.2x 1.9x 28.0% 23.4% 45.0% 0.7% 2.6% 3.1% 601318.SS Ping An Insurance Company Jiang, Jenny Insurance 97.0 87.0 722 211,424 10.2x 10.1x 2.4x 2.0x 28.0% 23.4% 45.0% 0.7% 2.4% 3.0% 2328.HK PICC P&C Company Ltd Jiang, Jenny Insurance 12.0 9.2 35 25,961 7.7x 7.8x 1.1x 1.0x 17.0% 14.3% 55.3% -1.0% 5.1% 5.0% 600585.SS Anhui Conch Cement Co. Ltd Zhang, Rachel Materials 49.0 41.3 135 30,792 6.8x 6.9x 1.6x 1.4x 28.8% 23.2% 8.8% -2.5% 4.4% 4.3% 0914.HK Anhui Conch Cement Co. Ltd Zhang, Rachel Materials 57.0 46.2 49 30,792 6.9x 7.1x 1.6x 1.4x 28.8% 23.2% 8.8% -2.5% 4.4% 4.3% 600019.SS Baoshan Iron & Steel Zhang, Rachel Materials 7.3 5.9 41 18,259 10.2x 9.2x 0.7x 0.7x 7.2% 7.7% (40.9%) 10.7% 4.9% 5.4% 2600.HK Aluminum Corp. of China Ltd. Zhang, Rachel Materials 3.1 2.4 8 7,575 16.8x 15.8x 0.7x 0.7x 4.2% 4.3% 154.7% 6.3% 0.0% 0.0% 601600.SS Aluminum Corp. of China Ltd. Zhang, Rachel Materials 3.9 3.5 39 7,575 26.7x 25.2x 1.1x 1.0x 4.2% 4.3% 154.7% 6.3% 0.0% 0.0% 0358.HK Jiangxi Copper Zhang, Rachel Materials 12.0 8.9 5 5,740 10.2x 9.1x 0.5x 0.5x 5.5% 5.9% 12.2% 11.5% 2.5% 2.8% 600362.SS Jiangxi Copper Zhang, Rachel Materials 14.0 14.4 37 5,740 18.1x 16.2x 1.0x 0.9x 5.5% 5.9% 12.2% 11.5% 1.4% 1.6% 0700.HK Tencent Holdings Ltd. Chen, Grace Media&Ent 430.0 322.8 823 389,704 28.8x 23.8x 7.2x 5.9x 30.3% 29.7% 24.4% 20.7% 0.4% 0.5% BIDU.O Baidu Inc Chen, Grace Media&Ent 132.0 101.5 534 35,466 48.3x 18.3x 1.5x 1.3x 3.2% 8.1% (80.9%) 164.4% 0.0% 0.0% 600276.SS Jiangsu Hengrui Wu, Sean Pharma 73.3 80.7 182 49,923 71.8x 54.5x 14.7x 11.8x 25.2% 27.0% 21.4% 31.8% 0.1% 0.2% 0016.HK Sun Hung Kai Properties Choudhary, Praveen Real Estate 132.0 110.9 74 40,991 11.8x 9.6x 0.7x 0.5x 6.0% 5.9% 6.6% 2.9% 3.7% 4.7% 1109.HK China Resources Land Ltd. Chen, Elly Real Estate 41.2 33.1 45 29,264 9.4x 8.1x 1.3x 1.2x 16.1% 16.5% 15.7% 15.5% 3.7% 4.3% 1113.HK CK Asset Holdings Ltd Choudhary, Praveen Real Estate 60.0 52.9 46 25,780 6.2x 9.1x 0.5x 0.5x 9.2% 5.9% 29.9% -31.6% 4.0% 4.4% 0960.HK Longfor Group Holdings Ltd. Chen, Elly Real Estate 32.4 29.6 22 22,385 9.9x 7.9x 1.7x 1.5x 19.9% 22.0% 26.4% 24.0% 4.6% 5.7% 1918.HK Sunac China Holdings Limited Chen, Elly Real Estate 55.7 32.3 83 18,126 4.6x 3.7x 1.6x 1.2x 49.6% 44.0% 31.3% 22.9% 4.8% 5.9% 1997.HK Wharf Real Estate Investment Company Ltd Choudhary, Praveen Real Estate 48.0 42.4 21 16,401 12.5x 12.0x 0.6x 0.6x 4.7% 4.8% 2.1% 4.7% 5.2% 5.4% 0101.HK Hang Lung Properties Ltd. Choudhary, Praveen Real Estate 20.0 17.7 13 10,133 17.0x 16.2x 0.6x 0.6x 3.4% 3.5% 14.0% 4.9% 4.3% 4.4% 0683.HK Kerry Properties Choudhary, Praveen Real Estate 32.0 24.3 7 4,515 6.1x 5.8x 0.3x 0.3x 5.9% 6.0% 72.1% 5.1% 5.6% 5.7% BABA.N Alibaba Group Holding Chen, Grace Retailing 207.0 168.3 609 439,315 36.7x 39.7x 6.5x 5.3x 23.9% 16.2% 36.2% -9.2% 0.0% 0.0% 3690.HK Meituan Dianping Chen, Grace Retailing 78.0 84.7 116 64,863 NM 71.7x 5.7x 5.4x (2.0%) 8.1% 80.3% 481.3% 0.0% 0.0% CTRP.O Ctrip.Com International Ltd Poon, Alex Retailing 35.0 30.0 176 20,866 33.0x 19.7x 1.5x 1.4x 4.9% 7.6% (5.4%) 67.1% 0.0% 0.0% 0780.HK Tongcheng-Elong Holdings Ltd Poon, Alex Retailing 17.0 12.3 8 3,327 15.4x 12.1x 1.9x 1.7x 6.2% 11.9% 32.6% 26.7% 0.0% 0.0% 2330.TW TSMC Chan, Charlie Semiconductors 288.0 286.5 265 240,531 22.0x 19.1x 4.9x 4.5x 20.1% 25.9% (3.9%) 15.3% 3.5% 3.8% 2454.TW MediaTek Chan, Charlie Semiconductors 449.0 384.5 66 19,915 25.1x 15.5x 2.1x 1.9x 8.8% 13.4% 16.0% 62.0% 3.0% 4.9% 603501.SS Will Semiconductor Co Ltd Shanghai Chan, Charlie Semiconductors 109.0 98.1 35 6,255 104.3x 37.5x 6.3x 5.5x 37.7% 16.8% 208.9% 178.1% 0.2% 0.9% 603986.SS GigaDevice Semiconductor Beijing Inc Yen, Daniel Semiconductors 179.0 145.4 83 5,868 75.4x 41.4x 16.5x 12.5x 26.7% 39.8% 35.1% 81.9% 0.5% 1.0% 0522.HK ASM Pacific Chan, Charlie Semiconductors 105.0 96.4 15 4,889 36.8x 15.0x 3.2x 3.0x 8.8% 21.0% (49.2%) 146.1% 1.9% 4.7% 3105.TWO WIN Semiconductors Corp Chan, Charlie Semiconductors 309.0 290.5 77 3,975 32.7x 22.3x 4.6x 4.1x 14.8% 20.7% 20.3% 46.5% 1.5% 1.8% 2379.TW Realtek Semiconductor Yen, Daniel Semiconductors 260.0 233.5 26 3,721 19.1x 17.4x 4.4x 4.1x 25.3% 24.9% 42.8% 9.5% 4.3% 4.7% 2337.TW Macronix International Co Ltd Yen, Daniel Semiconductors 35.0 32.7 39 1,908 19.3x 14.8x 1.8x 1.6x 9.9% 11.9% (65.7%) 30.3% 1.2% 1.6% 600588.SS Yonyou Network Technology Co Ltd Shih, Sharon Software & Svs 38.0 30.9 90 10,663 87.4x 69.0x 8.8x 8.3x 11.5% 12.8% 42.7% 26.6% 0.5% 0.7% GDS.O GDS Holdings Ltd Liu, Yang Software & Svs 48.0 40.9 32 5,709 NM NM 4.9x 4.9x (7.7%) (1.0%) 11.9% 81.5% 0.0% 0.0% 002410.SZ Glodon Co. Ltd. Liu, Yang Software & Svs 40.0 35.5 35 5,694 108.3x 57.2x 12.0x 10.5x 11.7% 21.1% (10.1%) 89.3% 0.6% 1.0% 002439.SZ VenusTech Liu, Yang Software & Svs 38.0 32.0 34 4,167 49.1x 37.5x 7.0x 6.0x 16.7% 19.0% 33.7% 31.0% 0.2% 0.3% 300383.SZ Beijing Sinnet Technology Liu, Yang Software & Svs 22.5 18.6 46 4,006 32.9x 24.8x 3.2x 2.7x 11.6% 12.9% 23.7% 32.6% 0.9% 1.2% VNET.O 21Vianet Liu, Yang Software & Svs 20.0 7.5 3 842 NM NM 1.2x 1.2x (1.3%) (0.2%) (0.2%) 85.8% 0.0% 0.0% 002415.SZ HIKVision Digital Technology Tsai, Yunchen Tech HW 35.0 32.3 221 41,388 23.5x 19.8x 6.8x 5.8x 33.9% 34.2% 10.9% 18.4% 2.1% 2.5% 601138.SS Foxconn Industrial Internet Co. Ltd. Shih, Sharon Tech HW 16.8 14.4 94 39,679 17.0x 15.6x 3.3x 2.8x 23.2% 21.2% (1.4%) 9.2% 0.9% 1.0% 000063.SZ ZTE Corporation Tsai, Yunchen Tech HW 34.0 32.0 445 17,615 26.2x 33.4x 3.6x 3.3x 15.5% 10.8% 173.0% -21.4% 1.3% 1.4% 0763.HK ZTE Corporation Tsai, Yunchen Tech HW 26.0 21.3 35 17,615 15.9x 20.2x 2.2x 2.0x 15.5% 10.8% 173.0% -21.4% 2.1% 2.3% 002281.SZ Accelink Technologies Co. Ltd. Tsai, Yunchen Tech HW 31.0 28.4 57 2,498 42.1x 30.9x 4.7x 4.3x 12.3% 15.4% 27.5% 36.1% 0.6% 0.8% 0788.HK China Tower Corp Ltd Yu, Gary Telecom Svs 2.5 1.8 119 39,740 51.4x 33.2x 1.5x 1.5x 3.1% 4.6% 75.8% 54.9% 1.2% 2.1% 0728.HK China Telecom Yu, Gary Telecom Svs 4.8 3.6 24 37,062 11.6x 10.3x 0.7x 0.7x 6.7% 7.2% 12.9% 11.7% 3.8% 4.0% 0762.HK China Unicom Yu, Gary Telecom Svs 12.0 8.4 36 32,630 17.3x 11.7x 0.7x 0.7x 4.3% 6.1% 48.7% 47.2% 2.5% 3.9% 002352.SZ S.F. Holding Co Ltd Fan, Qianlei Transportation 31.2 39.4 30 24,370 37.7x 29.7x 4.3x 3.9x 12.6% 14.6% 0.3% 26.9% 0.5% 0.7% 603056.SS Deppon Logistics Co Ltd Fan, Qianlei Transportation 15.7 13.2 8 1,776 19.4x 16.6x 2.9x 2.5x 16.3% 17.1% (6.4%) 16.4% 1.5% 1.8% 1816.HK CGN Power Co., Ltd Lee, Simon Utilities 1.8 1.9 9 11,131 8.7x 7.6x 1.0x 0.9x 12.8% 13.4% 4.7% 13.7% 3.7% 4.3%

Source: Modelware, Morgan Stanley Research. Data as of September 30, 2019 for A-shares, October 4, 2019 for HK listed names and October 7 for US listed names. *This table only summarizes potential beneficiaries that are covered by Morgan Stanley Research.

148 M BLUEPAPER Disclosure Section

The information and opinions in Morgan Stanley Research were prepared or are disseminated by Morgan Stanley Asia Limited (which accepts the responsibility for its contents) and/or Morgan Stanley Asia (Singapore) Pte. (Registration number 199206298Z) and/or Morgan Stanley Asia (Singapore) Securities Pte Ltd (Registration number 200008434H), regulated by the Monetary Authority of Singapore (which accepts legal responsibility for its contents and should be contacted with respect to any matters arising from, or in connection with, Morgan Stanley Research), and/or Morgan Stanley Taiwan Limited and/or Morgan Stanley & Co International plc, Seoul Branch, and/or Morgan Stanley Australia Limited (A.B.N. 67 003 734 576, holder of Australian financial services license No. 233742, which accepts responsibility for its contents), and/or Morgan Stanley Wealth Management Australia Pty Ltd (A.B.N. 19 009 145 555, holder of Australian financial services license No. 240813, which accepts responsibility for its contents), and/or Morgan Stanley India Company Private Limited, regulated by the Securities and Exchange Board of India (“SEBI”) and holder of licenses as a Research Analyst (SEBI Registration No. INH000001105); Stock Broker (BSE Registration No. INB011054237 and NSE Registration No. INB/INF231054231), Merchant Banker (SEBI Registration No. INM000011203), and depository participant with National Securities Depository Limited (SEBI Registration No. IN-DP-NSDL-372-2014) which accepts the responsibility for its contents and should be contacted with respect to any matters arising from, or in connection with, Morgan Stanley Research, and/or PT. Morgan Stanley Sekuritas Indonesia and their affiliates (collectively, "Morgan Stanley").

For important disclosures, stock price charts and equity rating histories regarding companies that are the subject of this report, please see the Morgan Stanley Research Disclosure Website at www.morganstanley.com/researchdisclosures, or contact your investment representative or Morgan Stanley Research at 1585 Broadway, (Attention: Research Management), New York, NY, 10036 USA.

For valuation methodology and risks associated with any recommendation, rating or price target referenced in this research report, please contact the Client Support Team as follows: US/Canada +1 800 303-2495; Hong Kong +852 2848-5999; Latin America +1 718 754-5444 (U.S.); London +44 (0)20-7425-8169; Singapore +65 6834-6860; Sydney +61 (0)2-9770-1505; Tokyo +81 (0)3-6836-9000. Alternatively you may contact your investment representative or Morgan Stanley Research at 1585 Broadway, (Attention: Research Management), New York, NY 10036 USA. Analyst Certification

The following analysts hereby certify that their views about the companies and their securities discussed in this report are accurately expressed and that they have not received and will not receive direct or indirect compensation in exchange for expressing specific recommendations or views in this report: Charlie Chan; Elly Chen; Grace Chen; Praveen K Choudhary; Qianlei Fan, CFA; Tim Hsiao; Jenny Jiang, CFA; Shawn Kim; Simon H.Y. Lee, CFA; Lillian Lou; Jack Lu; Kevin Luo, CFA; Sharon Shih; Laura Wang; Sean Wu; Richard Xu, CFA; Jack Yeung; Gary Yu; Rachel L Zhang; Sheng Zhong.

Unless otherwise stated, the individuals listed on the cover page of this report are research analysts. Global Research Conflict Management Policy

Morgan Stanley Research has been published in accordance with our conflict management policy, which is available at www.morganstanley.com/institutional/research/conflictpolicies. A Portuguese version of the policy can be found at www.morganstanley.com.br Important US Regulatory Disclosures on Subject Companies

The analyst or strategist (or a household member) identified below owns the following securities (or related derivatives): Praveen K Choudhary - Alibaba Group Holding(GDR); Laura Wang - New Oriental Education &Technology Group(GDR, common or preferred stock).

As of September 30, 2019, Morgan Stanley beneficially owned 1% or more of a class of common equity securities of the following companies covered in Morgan Stanley Research: 21Vianet, Alibaba Group Holding, Anhui Conch Cement Co. Ltd, Baidu Inc, China Communication Service Co Ltd, China CYTS Tours Holding Co Ltd, China Railway Construction, China Telecom, CRRC Corp Ltd, Ctrip.Com International Ltd, Dongfeng Motor Group, Goldwind, Jiangsu Hengrui, Jiangxi Copper, MediaTek, Meituan Dianping, New Oriental Education &Technology Group, TAL Education Group, Viomi Technology Co Ltd, WIN Semiconductors Corp, ZTE Corporation.

Within the last 12 months, Morgan Stanley managed or co-managed a public offering (or 144A offering) of securities of China National Building Material Company, GDS Holdings Ltd, Industrial Bank Co. Ltd., Longfor Group Holdings Ltd., New Oriental Education &Technology Group, Ping An Insurance Company, Sunac China Holdings Limited, Tencent Holdings Ltd., Tongcheng-Elong Holdings Ltd.

Within the last 12 months, Morgan Stanley has received compensation for investment banking services from Alibaba Group Holding, China National Building Material Company, China Tower Corp Ltd, Industrial Bank Co. Ltd., Longfor Group Holdings Ltd., Ping An Insurance Company, Sunac China Holdings Limited, Tencent Holdings Ltd..

In the next 3 months, Morgan Stanley expects to receive or intends to seek compensation for investment banking services from 21Vianet, Alibaba Group Holding, Aluminum Corp. of China Ltd., Anhui Conch Cement Co. Ltd, Baidu Inc, CGN Power Co., Ltd, China Communication Service Co Ltd, China Merchants Bank, China National Building Material Company, China Railway Construction, China Resources Land Ltd., China Telecom, China Tower Corp Ltd, China Unicom, CK Asset Holdings Ltd, CRRC Corp Ltd, Ctrip.Com International Ltd, Dongfeng Motor Group, Foxconn Industrial Internet Co. Ltd., Galaxy Entertainment, Goldwind, Guangzhou Automobile Group, Hang Lung Properties Ltd., Henderson Land, HIKVision Digital Technology, Huayu Automotive, Industrial Bank Co. Ltd., Jiangxi Copper, Kerry Properties, Longfor Group Holdings Ltd., MediaTek, Meituan Dianping, Midea Group Co Ltd., New Oriental Education &Technology Group, Nexteer Automotive Group, Ping An Bank, Ping An Insurance Company, S.F. Holding Co Ltd, SAIC Motor Corp. Ltd., Sino Land, Sun Hung Kai Properties, Sunac China Holdings Limited, TAL Education Group, Tencent Holdings Ltd., Tongcheng-Elong Holdings Ltd, Viomi Technology Co Ltd.

Within the last 12 months, Morgan Stanley has received compensation for products and services other than investment banking services from 21Vianet, Alibaba Group Holding, Baidu Inc, China Merchants Bank, China National Building Material Company, China Railway Construction, China Unicom, CRRC Corp Ltd, Ctrip.Com International Ltd, Henderson Land, Industrial Bank Co. Ltd., Longfor Group Holdings Ltd., MediaTek, Meituan Dianping, Midea Group Co Ltd., New Oriental Education &Technology Group, Ping An Bank, Ping An Insurance Company, Sun Hung Kai Properties, Sunac China Holdings Limited, Tencent Holdings Ltd., Tongcheng-Elong Holdings Ltd, Wharf Real Estate Investment Company Ltd, ZTO Express.

Within the last 12 months, Morgan Stanley has provided or is providing investment banking services to, or has an investment banking client relationship with, the following company: 21Vianet,

MORGAN STANLEY RESEARCH 149 M BLUEPAPER Alibaba Group Holding, Aluminum Corp. of China Ltd., Anhui Conch Cement Co. Ltd, Baidu Inc, CGN Power Co., Ltd, China Communication Service Co Ltd, China Merchants Bank, China National Building Material Company, China Resources Land Ltd., China Telecom, China Tower Corp Ltd, China Unicom, CK Asset Holdings Ltd, CRRC Corp Ltd, Ctrip.Com International Ltd, Dongfeng Motor Group, Foxconn Industrial Internet Co. Ltd., Galaxy Entertainment, GDS Holdings Ltd, Goldwind, Guangzhou Automobile Group, Hang Lung Properties Ltd., Henderson Land, HIKVision Digital Technology, Huayu Automotive, Industrial Bank Co. Ltd., Jiangxi Copper, Kerry Properties, Longfor Group Holdings Ltd., MediaTek, Meituan Dianping, Midea Group Co Ltd., New Oriental Education &Technology Group, Nexteer Automotive Group, Ping An Bank, Ping An Insurance Company, S.F. Holding Co Ltd, SAIC Motor Corp. Ltd., Sino Land, Sun Hung Kai Properties, Sunac China Holdings Limited, TAL Education Group, Tencent Holdings Ltd., Tongcheng-Elong Holdings Ltd, Viomi Technology Co Ltd.

Within the last 12 months, Morgan Stanley has either provided or is providing non-investment banking, securities-related services to and/or in the past has entered into an agreement to provide services or has a client relationship with the following company: 21Vianet, Alibaba Group Holding, Baidu Inc, China Merchants Bank, China National Building Material Company, China Railway Construction, China Unicom, CRRC Corp Ltd, Ctrip.Com International Ltd, Henderson Land, Industrial Bank Co. Ltd., Longfor Group Holdings Ltd., MediaTek, Meituan Dianping, Midea Group Co Ltd., New Oriental Education &Technology Group, PICC P&C Company Ltd, Ping An Bank, Ping An Insurance Company, Sun Hung Kai Properties, Sunac China Holdings Limited, Taiwan Semiconductor Manufacturing Co Lt, TAL Education Group, Tencent Holdings Ltd., Tongcheng-Elong Holdings Ltd, TSMC, Wharf Real Estate Investment Company Ltd, ZTO Express.

An employee, director or consultant of Morgan Stanley is a director of Industrial Bank Co. Ltd.. This person is not a research analyst or a member of a research analyst's household.

Morgan Stanley & Co. LLC makes a market in the securities of 21Vianet, Alibaba Group Holding, Baidu Inc, China Mobile Limited, China Unicom, Ctrip.Com International Ltd, New Oriental Education &Technology Group, Taiwan Semiconductor Manufacturing Co Lt, TAL Education Group, TSMC, Viomi Technology Co Ltd.

The equity research analysts or strategists principally responsible for the preparation of Morgan Stanley Research have received compensation based upon various factors, including quality of research, investor client feedback, stock picking, competitive factors, firm revenues and overall investment banking revenues. Equity Research analysts' or strategists' compensation is not linked to investment banking or capital markets transactions performed by Morgan Stanley or the profitability or revenues of particular trading desks.

Morgan Stanley and its affiliates do business that relates to companies/instruments covered in Morgan Stanley Research, including market making, providing liquidity, fund management, commercial banking, extension of credit, investment services and investment banking. Morgan Stanley sells to and buys from customers the securities/instruments of companies covered in Morgan Stanley Research on a principal basis. Morgan Stanley may have a position in the debt of the Company or instruments discussed in this report. Morgan Stanley trades or may trade as principal in the debt securities (or in related derivatives) that are the subject of the debt research report.

Certain disclosures listed above are also for compliance with applicable regulations in non-US jurisdictions. STOCK RATINGS

Morgan Stanley uses a relative rating system using terms such as Overweight, Equal-weight, Not-Rated or Underweight (see definitions below). Morgan Stanley does not assign ratings of Buy, Hold or Sell to the stocks we cover. Overweight, Equal-weight, Not-Rated and Underweight are not the equivalent of buy, hold and sell. Investors should carefully read the definitions of all ratings used in Morgan Stanley Research. In addition, since Morgan Stanley Research contains more complete information concerning the analyst's views, investors should carefully read Morgan Stanley Research, in its entirety, and not infer the contents from the rating alone. In any case, ratings (or research) should not be used or relied upon as investment advice. An investor's decision to buy or sell a stock should depend on individual circumstances (such as the investor's existing holdings) and other considerations. Global Stock Ratings Distribution

(as of September 30, 2019)

The Stock Ratings described below apply to Morgan Stanley's Fundamental Equity Research and do not apply to Debt Research produced by the Firm.

For disclosure purposes only (in accordance with NASD and NYSE requirements), we include the category headings of Buy, Hold, and Sell alongside our ratings of Overweight, Equal-weight, Not-Rated and Underweight. Morgan Stanley does not assign ratings of Buy, Hold or Sell to the stocks we cover. Overweight, Equal-weight, Not-Rated and Underweight are not the equivalent of buy, hold, and sell but represent recommended relative weightings (see definitions below). To satisfy regulatory requirements, we correspond Overweight, our most positive stock rating, with a buy recommendation; we correspond Equal-weight and Not-Rated to hold and Underweight to sell recommendations, respectively.

Other Material Investment Services Clients Coverage Universe Investment Banking Clients (IBC) (MISC) Stock Rating Count % of Total Count % of Total IBC % of Rating Category Count % of Total Other MISC Category Overweight/Buy 1155 37% 281 42% 24% 532 37% Equal-weight/Hold 1432 46% 319 47% 22% 678 47% Not-Rated/Hold 1 0% 0 0% 0% 1 0% Underweight/Sell 558 18% 76 11% 14% 224 16% Total 3,146 676 1435

Data include common stock and ADRs currently assigned ratings. Investment Banking Clients are companies from whom Morgan Stanley received investment banking compensation in the last 12 months. Due to rounding off of decimals, the percentages provided in the "% of total" column may not add up to exactly 100 percent.

150 M BLUEPAPER Analyst Stock Ratings

Overweight (O). The stock's total return is expected to exceed the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Equal-weight (E). The stock's total return is expected to be in line with the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Not-Rated (NR). Currently the analyst does not have adequate conviction about the stock's total return relative to the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Underweight (U). The stock's total return is expected to be below the average total return of the analyst's industry (or industry team's) coverage universe, on a risk-adjusted basis, over the next 12-18 months.

Unless otherwise specified, the time frame for price targets included in Morgan Stanley Research is 12 to 18 months. Analyst Industry Views

Attractive (A): The analyst expects the performance of his or her industry coverage universe over the next 12-18 months to be attractive vs. the relevant broad market benchmark, as indicated below.

In-Line (I): The analyst expects the performance of his or her industry coverage universe over the next 12-18 months to be in line with the relevant broad market benchmark, as indicated below.

Cautious (C): The analyst views the performance of his or her industry coverage universe over the next 12-18 months with caution vs. the relevant broad market benchmark, as indicated below.

Benchmarks for each region are as follows: North America - S&P 500; Latin America - relevant MSCI country index or MSCI Latin America Index; Europe - MSCI Europe; Japan - TOPIX; Asia - relevant MSCI country index or MSCI sub-regional index or MSCI AC Asia Pacific ex Japan Index. Important Disclosures for Morgan Stanley Smith Barney LLC Customers

Important disclosures regarding the relationship between the companies that are the subject of Morgan Stanley Research and Morgan Stanley Smith Barney LLC or Morgan Stanley or any of their affiliates, are available on the Morgan Stanley Wealth Management disclosure website at www.morganstanley.com/online/researchdisclosures. For Morgan Stanley specific disclosures, you may refer to www.morganstanley.com/researchdisclosures.

Each Morgan Stanley Equity Research report is reviewed and approved on behalf of Morgan Stanley Smith Barney LLC. This review and approval is conducted by the same person who reviews the Equity Research report on behalf of Morgan Stanley. This could create a conflict of interest. Other Important Disclosures

Morgan Stanley & Co. International PLC and its affiliates have a significant financial interest in the debt securities of Alibaba Group Holding, Baidu Inc, China Railway Construction, China Resources Land Ltd., CRRC Corp Ltd, Ctrip.Com International Ltd, Industrial Bank Co. Ltd., Longfor Group Holdings Ltd., Sands China Ltd., Sunac China Holdings Limited, Tencent Holdings Ltd..

Morgan Stanley Research policy is to update research reports as and when the Research Analyst and Research Management deem appropriate, based on developments with the issuer, the sector, or the market that may have a material impact on the research views or opinions stated therein. In addition, certain Research publications are intended to be updated on a regular periodic basis (weekly/monthly/quarterly/annual) and will ordinarily be updated with that frequency, unless the Research Analyst and Research Management determine that a different publication schedule is appropriate based on current conditions.

Morgan Stanley is not acting as a municipal advisor and the opinions or views contained herein are not intended to be, and do not constitute, advice within the meaning of Section 975 of the Dodd-Frank Wall Street Reform and Consumer Protection Act.

Morgan Stanley produces an equity research product called a "Tactical Idea." Views contained in a "Tactical Idea" on a particular stock may be contrary to the recommendations or views expressed in research on the same stock. This may be the result of differing time horizons, methodologies, market events, or other factors. For all research available on a particular stock, please contact your sales representative or go to Matrix at http://www.morganstanley.com/matrix.

Morgan Stanley Research is provided to our clients through our proprietary research portal on Matrix and also distributed electronically by Morgan Stanley to clients. Certain, but not all, Morgan Stanley Research products are also made available to clients through third-party vendors or redistributed to clients through alternate electronic means as a convenience. For access to all available Morgan Stanley Research, please contact your sales representative or go to Matrix at http://www.morganstanley.com/matrix.

Any access and/or use of Morgan Stanley Research is subject to Morgan Stanley's Terms of Use (http://www.morganstanley.com/terms.html). By accessing and/or using Morgan Stanley Research, you are indicating that you have read and agree to be bound by our Terms of Use (http://www.morganstanley.com/terms.html). In addition you consent to Morgan Stanley processing your personal data and using cookies in accordance with our Privacy Policy and our Global Cookies Policy (http://www.morganstanley.com/privacy_pledge.html), including for the purposes of setting your preferences and to collect readership data so that we can deliver better and more personalized service and products to you. To find out more information about how Morgan Stanley processes personal data, how we use cookies and how to reject cookies see our Privacy Policy and our Global Cookies Policy (http://www.morganstanley.com/privacy_pledge.html).

If you do not agree to our Terms of Use and/or if you do not wish to provide your consent to Morgan Stanley processing your personal data or using cookies please do not access our research.

Morgan Stanley Research does not provide individually tailored investment advice. Morgan Stanley Research has been prepared without regard to the circumstances and objectives of those who receive it. Morgan Stanley recommends that investors independently evaluate particular investments and strategies, and encourages investors to seek the advice of a financial adviser. The appropriateness of an investment or strategy will depend on an investor's circumstances and objectives. The securities, instruments, or strategies discussed in Morgan Stanley Research

MORGAN STANLEY RESEARCH 151 M BLUEPAPER may not be suitable for all investors, and certain investors may not be eligible to purchase or participate in some or all of them. Morgan Stanley Research is not an offer to buy or sell or the solicitation of an offer to buy or sell any security/instrument or to participate in any particular trading strategy. The value of and income from your investments may vary because of changes in interest rates, foreign exchange rates, default rates, prepayment rates, securities/instruments prices, market indexes, operational or financial conditions of companies or other factors. There may be time limitations on the exercise of options or other rights in securities/instruments transactions. Past performance is not necessarily a guide to future performance. Estimates of future performance are based on assumptions that may not be realized. If provided, and unless otherwise stated, the closing price on the cover page is that of the primary exchange for the subject company's securities/instruments.

The fixed income research analysts, strategists or economists principally responsible for the preparation of Morgan Stanley Research have received compensation based upon various factors, including quality, accuracy and value of research, firm profitability or revenues (which include fixed income trading and capital markets profitability or revenues), client feedback and competitive factors. Fixed Income Research analysts', strategists' or economists' compensation is not linked to investment banking or capital markets transactions performed by Morgan Stanley or the profitability or revenues of particular trading desks.

The "Important US Regulatory Disclosures on Subject Companies" section in Morgan Stanley Research lists all companies mentioned where Morgan Stanley owns 1% or more of a class of common equity securities of the companies. For all other companies mentioned in Morgan Stanley Research, Morgan Stanley may have an investment of less than 1% in securities/instruments or derivatives of securities/instruments of companies and may trade them in ways different from those discussed in Morgan Stanley Research. Employees of Morgan Stanley not involved in the preparation of Morgan Stanley Research may have investments in securities/instruments or derivatives of securities/instruments of companies mentioned and may trade them in ways different from those discussed in Morgan Stanley Research. Derivatives may be issued by Morgan Stanley or associated persons.

With the exception of information regarding Morgan Stanley, Morgan Stanley Research is based on public information. Morgan Stanley makes every effort to use reliable, comprehensive information, but we make no representation that it is accurate or complete. We have no obligation to tell you when opinions or information in Morgan Stanley Research change apart from when we intend to discontinue equity research coverage of a subject company. Facts and views presented in Morgan Stanley Research have not been reviewed by, and may not reflect information known to, professionals in other Morgan Stanley business areas, including investment banking personnel.

Morgan Stanley Research personnel may participate in company events such as site visits and are generally prohibited from accepting payment by the company of associated expenses unless pre-approved by authorized members of Research management.

Morgan Stanley may make investment decisions that are inconsistent with the recommendations or views in this report.

To our readers based in Taiwan or trading in Taiwan securities/instruments: Information on securities/instruments that trade in Taiwan is distributed by Morgan Stanley Taiwan Limited ("MSTL"). Such information is for your reference only. The reader should independently evaluate the investment risks and is solely responsible for their investment decisions. Morgan Stanley Research may not be distributed to the public media or quoted or used by the public media without the express written consent of Morgan Stanley. Any non-customer reader within the scope of Article 7-1 of the Taiwan Stock Exchange Recommendation Regulations accessing and/or receiving Morgan Stanley Research is not permitted to provide Morgan Stanley Research to any third party (including but not limited to related parties, affiliated companies and any other third parties) or engage in any activities regarding Morgan Stanley Research which may create or give the appearance of creating a conflict of interest. Information on securities/instruments that do not trade in Taiwan is for informational purposes only and is not to be construed as a recommendation or a solicitation to trade in such securities/instruments. MSTL may not execute transactions for clients in these securities/instruments.

Certain information in Morgan Stanley Research was sourced by employees of the Shanghai Representative Office of Morgan Stanley Asia Limited for the use of Morgan Stanley Asia Limited.

Morgan Stanley is not incorporated under PRC law and the research in relation to this report is conducted outside the PRC. Morgan Stanley Research does not constitute an offer to sell or the solicitation of an offer to buy any securities in the PRC. PRC investors shall have the relevant qualifications to invest in such securities and shall be responsible for obtaining all relevant approvals, licenses, verifications and/or registrations from the relevant governmental authorities themselves. Neither this report nor any part of it is intended as, or shall constitute, provision of any consultancy or advisory service of securities investment as defined under PRC law. Such information is provided for your reference only.

Morgan Stanley Research is disseminated in Brazil by Morgan Stanley C.T.V.M. S.A. located at Av. Brigadeiro Faria Lima, 3600, 6th floor, São Paulo - SP, Brazil; and is regulated by the Comissão de Valores Mobiliários; in Mexico by Morgan Stanley México, Casa de Bolsa, S.A. de C.V which is regulated by Comision Nacional Bancaria y de Valores. Paseo de los Tamarindos 90, Torre 1, Col. Bosques de las Lomas Floor 29, 05120 Mexico City; in Japan by Morgan Stanley MUFG Securities Co., Ltd. and, for Commodities related research reports only, Morgan Stanley Capital Group Japan Co., Ltd; in Hong Kong by Morgan Stanley Asia Limited (which accepts responsibility for its contents) and by Morgan Stanley Asia International Limited, Hong Kong Branch; in Singapore by Morgan Stanley Asia (Singapore) Pte. (Registration number 199206298Z) and/or Morgan Stanley Asia (Singapore) Securities Pte Ltd (Registration number 200008434H), regulated by the Monetary Authority of Singapore (which accepts legal responsibility for its contents and should be contacted with respect to any matters arising from, or in connection with, Morgan Stanley Research) and by Morgan Stanley Asia International Limited, Singapore Branch (Registration number T11FC0207F); in Australia to "wholesale clients" within the meaning of the Australian Corporations Act by Morgan Stanley Australia Limited A.B.N. 67 003 734 576, holder of Australian financial services license No. 233742, which accepts responsibility for its contents; in Australia to "wholesale clients" and "retail clients" within the meaning of the Australian Corporations Act by Morgan Stanley Wealth Management Australia Pty Ltd (A.B.N. 19 009 145 555, holder of Australian financial services license No. 240813, which accepts responsibility for its contents; in Korea by Morgan Stanley & Co International plc, Seoul Branch; in India by Morgan Stanley India Company Private Limited; in Indonesia by PT. Morgan Stanley Sekuritas Indonesia; in Canada by Morgan Stanley Canada Limited, which has approved of and takes responsibility for its contents in Canada; in Germany and the European Economic Area where required by Morgan Stanley Europe S.E., regulated by Bundesanstalt fuer Finanzdienstleistungsaufsicht (BaFin); in Spain by Morgan Stanley, S.V., S.A., a Morgan Stanley group company, which is supervised by the Spanish Securities Markets Commission (CNMV) and states that Morgan Stanley Research has been written and distributed in accordance with the rules of conduct applicable to financial research as established under Spanish regulations; in the US by Morgan Stanley & Co. LLC, which accepts responsibility for its contents. Morgan Stanley & Co. International plc, authorized by the Prudential Regulatory Authority and regulated by the Financial Conduct Authority and the Prudential Regulatory Authority, disseminates in the UK research that it has prepared, and approves solely for the purposes of section 21 of the Financial Services and Markets Act 2000, research which has been prepared by any of its affiliates. RMB Morgan Stanley Proprietary Limited is a member of the JSE Limited and A2X (Pty) Ltd. RMB Morgan Stanley Proprietary Limited is a joint venture owned equally by Morgan Stanley International Holdings Inc. and RMB Investment Advisory (Proprietary) Limited, which is wholly owned by FirstRand Limited. The information in Morgan Stanley Research is being disseminated by Morgan Stanley Saudi Arabia, regulated by the Capital Market Authority in the Kingdom of Saudi Arabia , and is directed at Sophisticated investors only.

Morgan Stanley Hong Kong Securities Limited is the liquidity provider/market maker for securities of Anhui Conch Cement Co. Ltd, China Merchants Bank, China Mobile Limited, China Resources

152 M BLUEPAPER Land Ltd., China Telecom, China Tower Corp Ltd, China Unicom, CK Asset Holdings Ltd, Galaxy Entertainment, Henderson Land, Jiangxi Copper, Meituan Dianping, PICC P&C Company Ltd, Ping An Insurance Company, Sands China Ltd., Sun Hung Kai Properties, Sunac China Holdings Limited, Tencent Holdings Ltd. listed on the Stock Exchange of Hong Kong Limited. An updated list can be found on HKEx website: http://www.hkex.com.hk.

The information in Morgan Stanley Research is being communicated by Morgan Stanley & Co. International plc (DIFC Branch), regulated by the Dubai Financial Services Authority (the DFSA), and is directed at Professional Clients only, as defined by the DFSA. The financial products or financial services to which this research relates will only be made available to a customer who we are satisfied meets the regulatory criteria to be a Professional Client.

The information in Morgan Stanley Research is being communicated by Morgan Stanley & Co. International plc (QFC Branch), regulated by the Qatar Financial Centre Regulatory Authority (the QFCRA), and is directed at business customers and market counterparties only and is not intended for Retail Customers as defined by the QFCRA.

As required by the Capital Markets Board of Turkey, investment information, comments and recommendations stated here, are not within the scope of investment advisory activity. Investment advisory service is provided exclusively to persons based on their risk and income preferences by the authorized firms. Comments and recommendations stated here are general in nature. These opinions may not fit to your financial status, risk and return preferences. For this reason, to make an investment decision by relying solely to this information stated here may not bring about outcomes that fit your expectations.

The trademarks and service marks contained in Morgan Stanley Research are the property of their respective owners. Third-party data providers make no warranties or representations relating to the accuracy, completeness, or timeliness of the data they provide and shall not have liability for any damages relating to such data. The Global Industry Classification Standard (GICS) was developed by and is the exclusive property of MSCI and S&P.

Morgan Stanley Research, or any portion thereof may not be reprinted, sold or redistributed without the written consent of Morgan Stanley.

Indicators and trackers referenced in Morgan Stanley Research may not be used as, or treated as, a benchmark under Regulation EU 2016/1011, or any other similar framework.

The Americas Europe Japan Asia/Pacific 1585 Broadway 20 Bank Street, Canary Wharf 1-9-7 Otemachi, Chiyoda-ku 1 Austin Road West New York, NY 10036-8293 London E14 4AD Tokyo 100-8104 Kowloon United States United Kingdom Japan Hong Kong Tel: +1 (1) 212 761 4000 Tel: +44 (0) 20 7 425 8000 Tel: +81 (0) 3 6836 5000 Tel: +852 2848 5200