Report No: AUS0001094 Realizing Armenia’s High-Tech Potential REALIZING ARMENIA'S HIGH-TECH POTENTIAL January 2020 © 2020 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Attribution —Please cite the work as follows: “World Bank. 2020. Realizing Armenia’s Technology Potential. Washington, D.C.: World Bank.” All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. CONTENTS ACKNOWLEDGMENTS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV EXECUTIVE SUMMARY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 CHAPTER 1 ARMENIA’S CURRENT POSITIONING IN HIGH-TECH GVCS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 CHAPTER 2 STRATEGIC GVC REPOSITIONING OPTIONS FOR ARMENIA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.1 Leveraging Armenia’s factor conditions to develop global niche activities 9 2.2 Upgrading to “adjacent” higher-value-added activities 12 2.3 Building capacity and relevance in nascent technologies while they are still accessible 19 CHAPTER 3 STRENGTHENING THE ECOSYSTEM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 CHAPTER 4 ACTIONS AND RECOMMENDATIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 APPENDIXES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 NOTES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Figures 1.1 ICT services exports vs. other high-tech exports, 2009–17  4 1.2 ICT services exports vs. total goods and services exports, 2009–17 5 1.3 ICT services exports per capita, Armenia and regional high performer comparators, 1995–2017 6 1.4 ICT exports per capita vs. CAGR of ICT services exports 7 1.5 Turnover of Armenian ICT firms by type 8 2.1 Market size through 2023 for 15 clean technologies in emerging economies 11 2.2 Size of the clean technology market accessible to SMEs in Eastern Europe 11 2.6 NSMS vs. ICT 13 2.7 Monthly average salaries of faculty in public universities in 28 countries, 2012  15 2.8 Capital expenditure requirements, firm consolidation, and barriers to entry are different in different segments 18 2.9 Citations of Armenian research papers in international journals, 2018 21 CS2.1 Male cancer mortality rates in 183 countries, 2016 24 CS2.2 Structure of the digital health GVC (at intersection with biotech and genomics) 26 CS2.3 Armenia’s participation in the biotech/digital health value chain 27 Tables ES.1 Actions and recommendations ix CS2.1 Health care expenditure per capita, 2016  22 4.1 Summary of recommendations  34 iii ACKNOWLEDGMENTS This report was written by Dube, Alejandro Espinosa-Wang, Ifeyinwa Onugha under the Lillyana Daza Jaller, Martin guidance of Wolfgang Fengler. Molinuevo, Arsen Nazaryan, Special thanks to Gohar Alanna Simpson, and Siegfried Malumyan for her data analysis Zottel. Thanks also to firms and and her insights and contributions individuals in Yerevan, Silicon relating to local trends. Thanks Valley, and Washington D.C. who also for contributions and insights were willing to participate in from World Bank Group consultations to aid the colleagues and associates, preparation of this report. including Wim Douw, Tafadzwa iv EXECUTIVE SUMMARY Service exports remain Armenia’s Armenia’s economy is heavily reliant on domestic demand, which is itself best strategy to achieve higher exports driven by remittances that support and create well-paid jobs. local construction and consumption. Armenia will need to shift to an export- led strategy if it is to move from middle- income to high-income status. Because Armenia is a landlocked country with comparatively high trading costs in physical goods, high-tech digital exports will continue to be key to Armenia’s growth. This report seeks to identify strategic opportunities for Armenia’s private sector in high-tech global value chains (GVCs) and also highlights reforms that could support Armenia’s evolution. The study builds on the FY18 Armenia Systematic Country Diagnostic, which identifies value-chain development as a core building block for sectoral growth. Although the report is targeted at policy makers in the Armenian government, sectoral development strategies rely heavily on strong private sector engagement and open public-private dialogue. Armenia’s tech community may therefore also take interest in this analysis. v and disruptive technologies. The government can ARMENIA’S CURRENT POSITIONING facilitate this transition by strategically repositioning IN HIGH-TECH GLOBAL VALUE Armenia in (a) activities that leverage Armenia’s factor CHAINS conditions, (b) ‘adjacent’ higher-value-added activities that build on the existing competencies of Armenian Armenia’s participation in information and firms, and (c) the development of nascent technologies communication technology (ICT) services GVCs has while they are still accessible. In addition, there are a resulted in promising export gains. ICT services number of areas where the regulatory ecosystem could exports doubled between 2009 and 2017, growing from be strengthened. These opportunities are discussed about US$94 million to US$212 million. Most of this below. A caveat, however: the sectoral examples export activity is in software programming, building on highlighted in the following sections are relevant at the a strong history of outsourcing services. Armenian ICT time of writing. As the tech industry rapidly evolves, firms are moving away from a service model and toward some of the presented examples may “expire,” giving product design. Several firms (e.g., Synopsys and way to new opportunities in other emerging Mentor Graphics) are already integrated into GVCs technologies. through foreign subsidiaries, and a vibrant high-tech start-up movement has produced high-profile firms STRATEGIC GVC REPOSITIONING that have achieved global success. As an example, OPPORTUNITIES FOR ARMENIA PicsArt, valued at US$250 million, has developed an image-editing and social-networking app that has been LEVERAGING ARMENIA’S FACTOR CONDITIONS downloaded 600 million times. Another mobile app TO DEVELOP GLOBAL NICHE SECTORS developer with its origins and development team based Armenian firms can target global niche vertical in Armenia, SoloLearn, has 1 million active monthly applications that leverage Armenia’s comparative users globally. advantages and, in so doing, attract foreign firms to While export growth is encouraging, Armenian Armenia, thus “localizing” the global market. firms may struggle to stay competitive in high-tech Armenia has several special characteristics (e.g., data, GVCs if they continue to target highly saturated, geology, population characteristics, geography, and mass-market vertical sectors. Armenian firms are industrial history) that make it particularly conducive competing globally to deliver mass-market software to developing solutions for certain vertical applications. and hardware solutions for a range of vertical applica- With the support of strategic policies and investments, tions. Examples include human resource management Armenian firms can build competence in these niche software development, lifestyle mobile applications, higher-value-added activities. In addition, the govern- banking software, social media software, and educa- ment can build on these factor conditions by tional training kits. Many of these solutions are implementing a targeted investment promotion disconnected from other high-tech activities in the strategy to attract global firms in these niche sectors to value chain, limiting opportunities for upgrading or Armenia. In so doing, the government could “bring the vertical integration. In addition, because the customer global market to Armenia,” reducing information base for many of these applications is often far from asymmetries and encouraging close cooperation and Armenia, generating new business has proven difficult. collaboration between global firms and local innovators Remaining competitive may be a challenge for and start-ups. Armenia as global competition from artificial intelli- Biotech solutions, cleantech and seismic engi- gence (AI)–enabled technologies (and lower-wage neering are examples of vertical applications with countries) increases. rich potential for Armenia’s software and hardware As high-tech GVCs evolve, the Armenian govern- development community. Armenia’s population has ment can strategically support the country’s distinctive health characteristics that could be valuable migration toward activities that are catalyzed— to global biotech development. As the government rather than threatened—by emerging technologies. codifies these health characteristics through its digital Armenia’s tech community is poised to move beyond health record system, Armenia’s software development software development toward deeper participation in community can build on these assets as global demand the development and commercialization of emerging for digital health care solutions widen to include a vi  |  Executive Summary variety of users such as physicians, insurance compa- and identify opportunities for joint research labs in nies, research centers and pharmacies. With an collaboration with leading international research appropriate regulatory framework, Armenia can institutions (in both the public and the private sector) position itself as a global hub for digital health, biotech in this field. development and bioscience research. As a second With a stronger AI research base, the govern- example, Armenia’s mining history has resulted in ment can implement policies to facilitate access to significant ongoing air and water pollution and land data and to collect and share both administrative contamination. The “cleantech” industry could mitigate and statistical data for AI engineering. Success in AI some of this environmental damage. Not only does the engineering (adapting and applying existing AI algo- global cleantech industry have very strong growth rithms to “real-world” applications) will be contingent potential, small and medium enterprises in middle- and on access to proprietary data. Gaining access to foreign low-income countries will have important roles to play private sector data will require Armenian firms to over the next decade. Rather than relying on the local comply with global information security and manage- 2 mining industry as a market, Armenia can become a ment provisions, such as the ISO/IEC 27000-series and concentrated research, development, and prototyping the European Union General Data Protection testbed for global cleantech firms. Finally, Armenia’s Regulation 2016/679. The government can implement geology could also lend itself to geographical informa- training programs for private firms to build compliance tion system development, seismic engineering, and and meet international standards. The government can disaster risk management technologies. go further to help plug the gap in available data by improving the accessibility of its own data. Where UPGRADING TO “ADJACENT” HIGHER-VALUE- regular data collection is already taking place, policies ADDED ACTIVITIES and practices differ significantly between ministries, GVC upgrading can come through vertical integra- and there is no clear policy or necessary infrastructure tion of value-chain activities that are adjacent to governing efficient data exchange in the govern- Armenia’s existing ICT services exports. Policy ment-to-government, government-to-business, and makers can examine the high-tech activities that government-to-consumer domains. The government Armenian firms already perform well and implement can consider standardizing the collection of both targeted programs to support vertical integration into administrative and statistical data in addition to “adjacent” higher-value-added activities. This report designing policies relating to the classification, format- examines opportunities for upgrading into (a) AI ting, storage, and exchange of these data. research and engineering (building on Armenia’s competence in mathematics and computer science) » SEMICONDUCTOR DESIGN and (b) semiconductor design (building on Armenia’s Upgrading to fabless semiconductor design and 1 strong foothold in electronic design automation). Intellectual Property development is an attractive prospect, but extremely high start-up costs, distance » AI RESEARCH AND ENGINEERING to market and Wassenaar Agreement export With investment in tertiary research, Armenia’s controls are holding Armenia back. Synopsys, a strong heritage in mathematics could position it as a global leader in electronic design automation (EDA), is global hub for pure AI research. Pure AI research (the Armenia’s largest ICT employer, with 800 employees. creation of brand-new algorithms) relies on advanced Considering Synopsys together with Mentor Graphics, mathematical techniques. Armenia has a strong global Armenia has a core capability in EDA software for comparative advantage in mathematics and natural semiconductor design. In cities with vibrant semicon- sciences—even more so than in ICT. Until recently, ductor ecosystems, relatively small start-up firms are commercial opportunities for mathematicians in able to spin off from larger firms to create their own 3 Armenia have been few, but AI research has created “fabless” semiconductor firms or design “intellectual high demand for mathematicians globally. Armenia property (IP) cores”—the building blocks of semicon- already has some nascent activity in AI, but a lack of ductors. The value-added associated with these tertiary training is leading to a brain drain and under- activities is extremely high, but start-up costs are far mining local investment opportunities. The higher than the venture capital currently available in government can build an AI PhD program in Armenia Armenia. Although Cisco Systems (through acquisition REALIZING ARMENIA’S TECHNOLOGY POTENTIAL  |  vii of U.S. firm Memoir Systems) has begun IP core and fabless semiconductor design, this important milestone STRENGTHENING THE ECOSYSTEM may be difficult to replicate without heavy foreign investment capital or an aggressive ramp-up in venture Innovation and entrepreneurship are constrained capital funding in Armenia. In addition, Wassenaar by limited investment capital; in addition, corporate Agreement export controls limit the extent to which law is poorly suited to early stage financing and international firms can collaborate with Armenian firms investment. There is only limited venture capital in semiconductor design due to Armenia’s categoriza- available in Armenia to support business innovation tion as a Group D country by the U.S. Department of and entrepreneurial endeavors. At the same time, FDI Commerce. Strong lobbying and intervention by the opportunities are constrained by a 1994 law that does Armenian government will be needed to re-assign in not adequately address current investment realities. As the grouping or seek blanket exemptions. an example, the law does not allow investors to choose international courts over local courts in case of BUILD CAPACITY AND RELEVANCE IN NASCENT arbitration. In addition, some aspects of Armenia’s TECHNOLOGIES WHILE THEY ARE STILL corporate law are poorly suited to early stage financing ACCESSIBLE and investment. As a result of these bottlenecks, start-ups in Armenia’s tech community typically » QUANTUM COMPUTING incorporate abroad to attract investment, limiting local As emerging technologies continue to disrupt GVC tax revenue and spillover effects. structures, quantum computing presents an oppor- Growth financing gaps could be bridged by tunity for Armenian firms to actively participate in introducing innovative FinTech products and evolving GVCs, rather than lagging behind. providing targeted financial capability advisory Quantum computing research is still in its infancy. It is services to micro, small, and medium enterprises true that only a handful of firms globally have the (MSMEs). Small firms in Armenia rely heavily on capital, knowledge, and convening power to construct internal resources and retained earnings for growth prototype quantum computers, outside of hardware finance. The limited use of bank financing is related to development the quantum research ecosystem is vast, the high cost of borrowing in Armenia. Promotion of and the global playing field is level. In addition, private FinTech innovations (e.g., mobile banking) and use of sector funding (rather than state-funded programs) agents could help to expand outreach to underserved now dominates quantum research, opening the door to segments and locations. Expanding targeted financial researchers in middle- and low-income countries. capability advisory services to MSMEs on financial Global citations of Armenia’s research and develop- products, financial reporting, financial management, ment (R&D) in physics far exceed that of computer business planning and marketing would enhance their sciences and even mathematics, placing Armenia in the bankability and discourage informality, which is already top 25 countries in the world for R&D citations in at a very high level in the enterprise sector. nuclear physics. However, Armenia has no national Armenia already has a fairly strong framework quantum strategy and is not leveraging its potential in the area of electronic transactions and data comparative advantage in this area. governance, but these laws could be strengthened As the global race intensifies, quantum comput- and complemented by a framework to address ing strategy and road map will be increasingly intermediary liability. The government could important as will the need to invest in tertiary consider expanding the Law on Electronic Document education and research. The Armenian government and Electronic Signature to recognize all electronic could consider policies to encourage international signatures and could, in addition, grant them collaboration in the field of quantum research (e.g., “presumption of validity and authenticity” in legal applied research grants, competitive research collabo- proceedings. Armenia already has comprehensive rations, competitions, and R&D tax credits) and legislation to protect online personal data privacy. position Armenia as a regular “stop” on the global However, the Agency for the Protection of Personal quantum computing conference circuit. This should Data could consider including specific regulations for include formal tertiary training at all three levels— cybersecurity, adding requirements such as the bachelor’s, master’s, and doctoral. appointment of a data manager and risk assessment viii  |  Executive Summary procedures. Finally, the government could consider a framework that explicitly addresses intermediary NOTES liability, with safe harbor provisions to shield firms 1. Electronic design automation software from liability for infringing content posted by third comprises applications used by the semiconductor parties, including intellectual property violations. industry to design chips. Armenia’s potential in high-tech sectors is closely tied with its R&D development landscape and 2. ISO stands for International Organization tertiary education framework; both are in need of for Standardization. IEC stands for International targeted reforms. An increase of public funding for Electrotechnical Commission. R&D could support targeted research policies, includ- 3. “Fabless” semiconductor firms design ing applied research grants, competitive research and sell complete chipsets but outsource their collaborations, competitions, and R&D tax credits. At fabrication. the same, tertiary courses in engineering and computer science are in urgent need of overhaul. Additionally, universities can better work with the private sector to provide more professional development and on-the-job training options that open up mid-career learning options. TABLE ES.1 Actions and recommendations RECOMMENDATION AUTHORITY TIMEFRAME Entrepreneurship, research, and innovation » Develop an innovation policy road map that identifies policies MoESCS, MoHTI Long term that encourage international research collaboration in priority fields. » Provide support for research and innovation programs through MoESCS, Short term sub-financing, e.g., applied research grants, competitive MoHTI research collaborations and matching grants programs for start-ups. » Strengthen human resources, research excellence and research MoESCS Short term commercialization through government-facilitated collaborative research programs with international universities, private sector, and a targeted visa program. » Develop capacity of technology transfer offices to convert early MoESCS, Long term stage research to market-ready technology and complete deals. MoHTI » Support development of local and diaspora-led business angel MoHTI, MoE Long term networks in Armenia. » Shift focus of publicly funded seed-funding programs (including MoHTI Short term incubators and accelerators) toward “successful exits” rather than “quantity of firms” in the program. Table continues next page REALIZING ARMENIA’S TECHNOLOGY POTENTIAL  |  ix TABLE ES.1 (Continued) RECOMMENDATION AUTHORITY TIMEFRAME » Introduce software patenting legislation to protect software MoHTI, MoJ Long term that demonstrates “further technical effect” (in line with EU guidelines). » Introduce managerial training programs to build middle- MoE, MoHTI management capacity. Access to growth finance » Launch consultations and develop a fintech strategy to MoF, CBA Short term introduce innovative products that could help bridge the MSME financing gap. » Develop implementation strategy (including adequate MoF, CBA Short term regulatory safeguards and oversight) to use agents (e.g., the Armenian Post) for delivery of core financial services. » Expand targeted financial capability advisory services to MoF, CBA Long term MSMEs on financial products, financial reporting, financial management, business planning and marketing to enhance their bankability and discourage informality. Corporate law and foreign direct investment » Introduce debt instruments (e.g., convertible loans) that can be SRLE, MoE, MoF Short term converted to equity in subsequent funding rounds. » Introduce corporate entities that support different classes of SRLE, MoE Short term voting shares (with differing entitlements to dividends and/or different levels of voting power) in order to better reflect early stage financing risks. » Develop clear rules pertaining to the fiduciary responsibilities of SRLE, MoE Long term corporate management teams toward minority shareholders (so that the latter are adequately protected—particularly where the interests of minority shareholders directly clash with those of majority shareholders). » Introduce legislation that allows investors to choose between MoE, MoF Short term local courts and international courts in the event of arbitration. » Strengthen investment “aftercare” by creating clear MoE, MoF Short term mechanisms to identify, track, and manage grievances between investors and public agencies before they escalate into disputes or withdrawals. Table continues next page x  |  Executive Summary TABLE ES.1 (Continued) RECOMMENDATION AUTHORITY TIMEFRAME » Repeal 1994 Foreign Investment Law and enact new umbrella MoE, MoF Long term framework for investment in line with international best practice, including: › Defining scope of application of the “Fair and Equitable Treatment” principle, › Defining characteristics of compensation payment, › Defining limits of prohibition of direct and indirect expropriation, and › Proscribing the rights of the investor to dispose of their profits and export their funds. » Build “investor roadmap” to identify permits, processes and MoE, MoF Short term administrative changes that are constraining investments. » Design investment promotion strategy to attract global firms MoHTI, MoE, MoF Short term in niche sectors that can leverage Armenia’s particular factor conditions (e.g., biotech and cleantech). Digital trade » Expand the Law on Electronic Document and Electronic MoJ Long term Signature to recognize all electronic signatures and grant them, in addition, “presumption of validity and authenticity” in legal proceedings. » Include specific regulations for cybersecurity in the Law on MoJ Long term Personal Data Protection, adding requirements such as the appointment of a data manager and risk assessment procedures. » Develop framework that explicitly addresses online MoJ Long term intermediaries, with safe harbor provisions to shield them from liability for infringing content posted by third parties, including intellectual property violation. » Work with the U.S. Department of Commerce and international MoE Long term counterparts to improve Armenia’s grouping in order to facilitate trade and collaboration in “dual-use” technologies as per the international Wassenaar Agreement. Data policy     » Implement policies relating to the classification, formatting, and MoHTI, MoE Short term storage of administrative and statistical data. » Implement policies that will enhance digital data management MoHTI, MoE Short term capacity and promote digitization among private firms. Table continues next page REALIZING ARMENIA’S TECHNOLOGY POTENTIAL  |  xi TABLE ES.1 (Continued) RECOMMENDATION AUTHORITY TIMEFRAME » Implement localization and data sovereignty laws that do not MoHTI, MoE Long term stifle interest but still stimulate local economic activity in Armenia. » Provide technical assistance to help the private sector comply MoHTI, MoE Long term with the EU’s General Data Protection Regulations and meet international standards (e.g., the ISO/IEC 27000 family of standards). Education     » Integrate planned “High-Tech Industries Strategy” with planned MoESCS, MoHTI Short term “Education, Science and Culture, and Sports Strategy.” » Update bachelor’s, master’s, and PhD curricula in computer MoESCS, MoHTI Short term science and engineering fields to correspond with industry standards. » Introduce annual curriculum review process that uses public- MoESCS, MoHTI Short term private dialogue process to solicit industry opinion and incorporates clear mechanism for revisions and updates. » Build an AI PhD program and identify opportunities for joint MoESCS, MoHTI Short term research labs in collaboration with leading international research institutions (in both the public and the private sector) in this field. » Develop a quantum computing strategy and road map. This MoESCS, MoHTI Long term should include formal tertiary training in quantum computing at all three levels—bachelor’s, master’s, and doctoral. » Strengthen future incarnations of the “Programme for MoF, MoE, MoESCS Short term Anti-Corruption Measures in Education” as follows: › Provide adequate funding for coordination, monitoring, and implementation of all listed anti-corruption measures, › Reduce prevalence of short-term, noncompetitive appointments to increase employment security and predictability among university staff, › Introduce compliance and integrity into accreditation and reaccreditation criteria for higher education providers, and › Develop comprehensive corruption detection and reporting mechanisms in the higher education sector, in such a way as to protect whistleblowers. Note: MoESCS = Ministry of Education, Science, Culture, and Sports; MoHTI = Ministry of High-Tech Sectors; MoE = Ministry of Economy; MoF = Ministry of Finance; SRLE = State Register of Legal Entities, Ministry of Justice of the Republic of Armenia; MoJ = Ministry of Justice; EU =European Union; MSME = micro, small, and medium enterprise; CBA = Central Bank of Armenia. xii  |  Executive Summary ABBREVIATIONS AGP Armenian Genome Project AI artificial intelligence ArMed Armenia National Electronic Health Information System CBA Central Bank of Armenia DNA deoxyribonucleic Acid EDA electronic design automation EIF Enterprise Incubator Foundation EHR electronic health records FDI foreign direct investment FPGAs field programmable gate arrays GDP gross domestic product GIS geographical information system GVC global value chain ICT information and communication technology IEC International Electrotechnical Commission IoT internet of things IP intellectual property ISO International Organization for Standardization ISTC Innovative solutions and Technologies Center MIPT National Research University of Electronic Technology MoE Ministry of Economy MoESCS Ministry of Education, Science, Culture, and Sports MoF Ministry of Finance MoHTI Ministry of High-Tech Sectors MSME micro, small, and medium enterprise NPUA National Polytechnic University of Armenia NSMS Natural Sciences, Mathematics and Statistics NSSP National Service for Seismic Protection ACN Anti-Corruption Network for Eastern Europe and Central Asia R&D research and development SME small and medium enterprises SRLE State Register of Legal Entities, Ministry of Justice of the Republic of Armenia WA Wassenaar Arrangement REALIZING ARMENIA’S TECHNOLOGY POTENTIAL  |  xiii INTRODUCTION Armenia is poised to catalyze growth in high-tech consumer preferences. To the extent that the internet sectors. In the spring of 2018, large, peaceful nation- of things (IoT), advanced robotics, 3D printing, and wide street protests (referred to as Armenia’s “velvet other digital technologies associated with Industry 4.0 revolution”) resulted in new leadership, which went are labor-saving and intensify the “servicification” of on to consolidate a majority following parliamentary manufacturing, they render high-skill (and often elections in December 2018. With the new govern- high-wage) economies more attractive for manufactur- ment’s commitment to regulatory reform, there is an ing. In addition, the platform economy matches buyers opportunity to create a much stronger, start-up- and suppliers from all over the world in new delivery friendly business environment that better enables the models to create new markets. Armenia needs to be private sector to increase its productivity, accelerate ready to adapt and leverage these new digital technolo- growth, and attract talent and foreign direct invest- gies and upskill its workforce or else risk falling behind. ment into the country. In addition, a new Ministry of The purposes of this report are (a) to help policy High-Tech Industry has been created to design and makers identify strategic opportunities for implement policies and programs that will Armenia’s private sector in high-tech GVCs as they catalyze growth. evolve and (b) to highlight reforms that could Emerging digital technologies are changing the support this repositioning. As digital technologies structure of global value chains (GVCs) in high-tech continue to influence the structure of high-tech GVCs sectors. With the world now undergoing a fourth and as lower-wage countries continue to increase their industrial revolution (Industry 4.0), Armenia needs to market share, Armenia’s policy makers and private better position itself to leverage global digital trends. sector alike must be more strategic about GVC The World Development Report 2020: Trading in the Age of development and upgrading. This report seeks to Global Value Chains highlights the impact that techno- identify strategic opportunities for Armenia’s private logical progress has on GVC participation, including sector in high-tech GVCs and also highlights reforms reducing trading costs, inducing quality upgrading and that could support Armenia’s evolution. This study churn, and shifting relative productivity across sectors. builds on the fiscal 2018 Armenia Systematic Country The data economy can affect the geography of produc- Diagnostic (World Bank 2017), which identifies tion by reducing the importance of wage costs, value-chain development as a core building block for changing delivery models, and responding to changing sectoral growth. 2 The report uses a qualitative analytical dialogue with Armenia’s high-tech community may also approach1 to inform policy design and help the take an interest in this analysis. Armenian government identify strategic options to This report consists of four chapters. Chapter 1 retain a greater share of value added from global describes Armenia’s current position in high-tech industries. The approach uses primary research GVCs. Chapter 2 focuses on strategic repositioning and methods and interviews to identify the strategic upgrading options for Armenia. Chapter 3 describes position of Armenia’s high-tech firms in the GVCs they opportunities to strengthen Armenia’s regulatory serve. In particular, the approach appraises the ecosystem. Lastly, chapter 4 describes recommended capacity of Armenian firms to: (a) create value in the actions for strengthening Armenia’s participation in GVCs in which they participate, (b) guard their high-tech GVCs through policy reforms, capital competitive positions, and (c) upgrade to higher-val- investments, and skills development. ue-added activities. Consultations included almost 50 private firms, investors, and learning institutions from Yerevan and from San Diego, Silicon Valley, and Washington D.C. in the U.S. Armenia’s performance in identified segments was assessed compared with regional, as well as global, competitors, and the “ideal value chain” for those segments was compared with the Armenian context. These sectoral development strategies will rely heavily on strong engagement with Armenia’s private sector high-tech community. Successful design and implementation of sector-specific strategies relies heavily on intensive interaction between project teams, private sector beneficiaries and cluster organiza- tions. Any recommendations arising from this report should only be implemented through enhanced stakeholder involvement and facilitated public-private 3  | Introduction Chapter 1 ARMENIA’S CURRENT POSITIONING IN HIGH-TECH GVCS While information and communication technology US$150 to US$250 in ICT services per capita in 2017, (ICT) services dominate armenia’s high-tech export compared with Armenia’s US$72 in ICT services per activity, it is still a relatively small share of the capita. (Exports per capita of global leaders are far overall economy. Armenia’s international trade in higher, in excess of US$1,000 per capita in 2017.) high-tech sectors is dominated by ICT services, which far outweigh tangible ICT goods exports (see figure 1.1). Exports in ICT services more than doubled between FIGURE 1.1 2009 and 2017, growing from about US$94 million to ICT services exports vs. other high-tech US$212 million. In contrast, other high-tech exports exports, 2009–17 — (e.g., aerospace, pharmaceuticals, scientific instru- US$ millions ments, and electrical machinery) reached only US$26 million in 2017, and other ICT goods exports reached only US$4 million. Although growth in Armenia’s ICT services exports is encouraging, these exports comprised only 11 percent of total service exports and only 4.8 percent of goods and services exports in 2017 (see figure 1.2). Growth in ICT exports compares favorably with regional competitors, but there is still room for improvement. Armenia’s export performance compares favorably with some of its regional competi- ICT services exports ICT goods exports tors, including Georgia, Kazakhstan, the Russian Other high tech exports Federation, and Ukraine (see figure 1.3). In addition, Source: World Development Indicators database. Note: ICT = information and communication technology. ICT services exports include Armenia outperforms global ICT services giants China computer and communications services (telecommunications and postal and courier services) and information services (computer data, news-related service transactions). and India on a per capita basis (see figure 1.4). ICT goods exports include hardware such as computers and peripheral equipment, However, there is still room for improvement; Belarus, communication equipment, consumer electronic equipment, electronic components, and other (miscellaneous) information and technology goods. Other high-tech exports are Bulgaria, Hungary, Lithuania, Poland, Romania, products with high research and development intensity, such as in aerospace, computers, pharmaceuticals, scientific instruments, and electrical machinery. Slovenia, and the Slovak Republic exported between 4 strong source of growth, creating jobs, and technology FIGURE 1.2 ICT services exports vs. total goods and and skills transfer (including managerial and soft services exports, 2009–17 skills). However, the value created in Armenia through — these subsidiaries is not retained beyond salaries and US$ billions overhead costs. In recent years, a vibrant tech start-up culture has emerged, producing high-profile firms that have achieved global success. In the past few years, Armenia’s tech start-up ecosystem has begun to flourish. Up to 200 start-ups have emerged that are supported by programs and organizations providing idea validation, acceleration services, seed investments, networking and coworking facilitation, business centers, government initiatives, and entrepreneur associations. As an example, PicsArt, valued at US$250 million, has developed an image-editing and social-net- working app that has been downloaded 600 million times. Another mobile app developer with its origins Goods and services and development team based in Armenia, SoloLearn, Services only BOP has 1 million active monthly users globally. ICT services BOP Source: World Development Indicators database. Armenian firms are competing globally to deliver Note: BOP = balance of payments; ICT = information and communication technology. mass-market software and hardware solutions for a range of vertical applications. Armenia’s software development community is designing solutions for a Software programming dominates Armenia’s wide range of global vertical applications. Firms have ICT services exports, but it has historically targeted these activities and applications because they comprised outsourcing activities that are now have low barriers to entry and solutions that can be trending toward automation. Armenia initially grew readily developed and delivered remotely. Examples through low-cost outsourcing, notably customized include human resources management software software and web design (see figure 1.5). As noted in development, lifestyle mobile applications, banking the World Development Report 2020: Trading in the software, social media software, and educational age of Global Value Chains, global trends are toward a training kits. However, the same low barriers to entry reduction in reliance on human resources to perform that make the mass-market applications “attractive” such functions and, even within Armenia, several firms also make them highly competitive globally. have started to develop programs that automate In other cases, activities that Armenian firms software design services. As an example, the Armenian can compete for are (necessarily) disconnected from firm 10Web has developed a software that enables users other high-tech activities in the global value chain, to build WordPress websites rather than outsourcing limiting opportunities for upgrading or vertical design. Similarly, Renderforest, another Armenian integration. In a number of vertical applications, the company, uses artificial intelligence (AI) algorithms for reason that global firms are able to provide services animated video production. “remotely” is because the activity is relatively discon- Encouragingly, much of Armenia’s “product and nected from the rest of the value chain. This service solutions” activity (as opposed to outsourc- characteristic lowers the barriers to entry for that ing services) is already integrated into GVCs. Several specific activity but makes it difficult for Armenian global tech firms have operations in Armenia, with 38 firms to upgrade or vertically integrate “adjacent” percent of Armenia’s ICT turnover coming from higher-value-added activities. In some cases, there are subsidiaries of foreign firms. The most high-profile of few other high-tech activities in the value chain to these are Synopsys, Mentor Graphics, and National target for upgrading. For instance, several Armenian Instruments, which together employ about 1,200 firms design and manufacture standalone educational people in Armenia. This subsidiary model has been a electronics toolkits for schools and universities. While 5  |  Armenia’s Current Positioning in High-Tech GVCs FIGURE 1.3 ICT services exports per capita, Armenia and regional high performer comparators, 1995–2017 — a. Armenia and regional comparators, US$ Lithuania Poland Bulgaria Belarus Armenia Ukraine Russian Federation Georgia Kazakhstan — b. Armenia and High performer comparators, US$ Israel Switzerland Sweden Netherlands Estonia Germany Czech Republic Slovenia Hungary Romania Armenia Source: World Development Indicators database. 6  |  Armenia’s Current Positioning in High-Tech GVCs FIGURE 1.4 ICT exports per capita vs. CAGR of ICT services exports Average Israel Singapore Cyprus Switzerland Sweden Finland Netherlands Estonia United Kingdom Germany Norway Czech Republic France Slovenia Costa Rica Slovak Republic ICT services exports per capita BOP current US Hungary Romania Canada Malta Average Lithuania Croatia Poland Bulgaria Portugal Belarus Italy United States New Zealand Australia Uruguay Greece Mauritius Malaysia Korea Rep Panama Armenia Bosnia and Herzegovina Ukraine India Kyrgyz Republic Moldova Philippines Argentina Morocco Japan Sri Lanka Albania El Salvador Nicaragua Oman Russian Federation Honduras Azerbaijan Egypt Arab Rep Georgia Chile Thailand Tunisia China Guatemala Ethiopia Colombia Mongolia Namibia Pakistan Cambodia Indonesia Cameroon Kenya Peru Brazil Paraguay Bolivia Kazakhstan Lesotho Bangladesh Botswana ICT services exports’ nominal average CAGR % Source: World Development Indicators database. Note: BOP = balance of payments; CAGR = compound annual growth rate. Regional comparator countries are highlighted in blue. 7  |  Armenia’s Current Positioning in High-Tech GVCs technologies (see section 2.2). At the same time, the FIGURE 1.5 Turnover of Armenian ICT firms by type World Development Report 2020: Trading for Development in the Age of Global Value Chains notes that developing Armenian products Subsidiaries of and service solutions foreign firms countries stand the most to gain from the reduction in trade costs that arise from digital technologies (because they face the highest trade costs and biggest distor- tions), adding further competitive pressures. % The government can strategically support Armenia’s migration toward activities that are catalyzed—rather than threatened—by emerging % technologies. Armenia’s tech community is poised to % move beyond software development toward deeper participation in the development and commercializa- tion of emerging and disruptive technologies. As discussed in chapter 2, the government can facilitate % this transition by strategically repositioning Armenia’s ICT services providers in (a) activities that leverage Outsourcing and Armenia’s factor conditions (see section 2.1); (b) Infrastructure and software development “adjacent” high-value-added technologies that build on telecom services services the existing competencies of Armenian firms (see Source: Enterprise Incubator Foundation 2019. section 2.2); and (c) the development of nascent Note: ICT = information and communication technology. technologies while they are still accessible (see section 2.3). In addition, the regulatory ecosystem could be further strengthened (see chapter 3). These opportuni- these firms were able to enter the market relatively ties are discussed in the chapters that follow. 2 easily, they have found few opportunities to upgrade or expand their activities. In addition, as the customer base for many of these vertical applications is far from Armenia, generating new business has proven difficult. Whether for business-to-business or business-to-con- sumer solutions, the broad base of end-users for Armenian software products and solutions is outside of Armenia. Indeed, there is a heavy bias toward U.S. markets and clientele. Entry into certain sectors has often been serendipitous, and several Armenian start-ups and small and medium enterprises have indicated that they have struggled to generate new business and expand their customer base. At the same time, many of Armenia’s competitors are much closer to the end-user and may have a more intuitive under- standing of their needs. Continuing to compete for mass-market solutions will be increasingly difficult for Armenia as global competition from automation (and lower-wage countries) increase. While it is important to increase the value of Armenia’s ICT services exports, the types of exports (and underlying activities therein) are equally important, particularly as some software development functions face threats from AI and machine-learning 8  |  Armenia’s Current Positioning in High-Tech GVCs Chapter 2 STRATEGIC GVC REPOSITIONING OPTIONS FOR ARMENIA “High-tech sectors” is a vast grouping that in 2019 vant at the time of writing. As the technology revolu- could describe virtually all vertical sectors. While tion continues, some of the presented examples may discussions surrounding disruptive technology quickly “expire,” giving way to new opportunities in other turn to 3D printing, augmented reality, internet of emerging technologies. things (IoT), blockchain, and artificial intelligence (AI), the so-called “traditional” sectors (e.g., automotive and 2.1 pharmaceutical) are just as “digital” as Amazon. A widely shared 2017 infographic demonstrated that there LEVERAGING ARMENIA’S FACTOR were more lines of code in a modern high-end car (100 CONDITIONS TO DEVELOP GLOBAL NICHE million) than in Facebook (61 million) (McCandless ACTIVITIES 2017). Even agriculture is no longer exempt; global positioning satellite guidance systems, sensors, robot- Armenia’s software and hardware development ics, control systems, farm management software, community currently competes to deliver drones, and telematics are all frequently used to mass-market applications. Currently, Armenia’s increase the quantity and accuracy of information avail- software and hardware development community able to farmers in order to maximize returns, increase targets highly saturated global vertical sectors where: efficiency of production, and reduce waste (Schmaltz (a) Armenian firms have little or no comparative 2017). Crucially, each one of these vertical sectors has advantage, (b) opportunities for upgrading and vertical its own value chain and its own determinants of integration are limited, and (c) building a customer competitiveness. This chapter identifies different strat- base is reliant on proximity. Examples include human egies that could be employed by the government to resources management software development, lifestyle position Armenia to deliver higher-value-added, mobile applications, banking software, social media tech-related activities in identified sectors. As sectoral software, and educational training kits. development strategies rely heavily on strong private Armenian firms may do better to leverage sector engagement and open public-private dialogue, armenia’s factor conditions to target global niche Armenia’s tech community may also take interest in vertical applications. Armenia has several special this analysis. It is important to note that the sectoral characteristics (e.g., geology, population characteristics, examples highlighted in the following sections are rele- geography, and industrial history) that could make it 9 particularly conducive to developing solutions for With an appropriate regulatory framework, certain vertical applications. With the support of Armenia can position itself as a global hub for strategic policies and investments, there is an opportu- digital health, biotech development, and bioscience nity for Armenian firms to target these niches and build research. Building on these factor conditions, stronger offerings to better compete in global markets. Armenian firms could expand their global value chain This chapter describes three such global niches that (GVC) participation by moving from digital health to Armenia could develop. biotech devices and associated software solutions. This A strategic investment promotion strategy could could include (hardware and software) design of attract global firms in these niches, “localizing” the implantable devices (e.g., cardioverter defibrillators), global market. Where there is a strong “local” compo- wearable devices (e.g., continuous glucose monitors) nent to Armenia’s comparative advantage in a and IoT solutions (e.g., sensors for smart watches and particular niche (for instance, those related to geology mobile applications). In so doing, Armenia could or population characteristics), it might be advanta- become a global hub for digital health technologies, geous for foreign firms to locate in Armenia to take attracting foreign direct investment and international advantage of certain factor conditions for research, research institutions. (This opportunity for Armenia is development and testing. The Armenian government discussed in depth in the case study on ICT services for has the opportunity to orient investment promotion digital health and biotech applications, page 22.) policies toward these niches. In so doing, the govern- ment could “bring the global market to Armenia,” CLEANTECH reducing information asymmetries and encouraging Armenia’s mining history has resulted in significant close cooperation and collaboration between global ongoing air and water pollution and land contami- firms and local innovators and start-ups. nation. Mining is Armenia’s dominant export sector. Copper ore alone accounted for 35 percent of Armenia’s BIOTECH AND DIGITAL HEALTH exports in 2017, while aluminum foil, raw copper, gold, Armenia’s population has distinctive health charac- and diamonds accounted for 16.8 percent. teristics that could be valuable to global biotech Unfortunately, mining practices continued from the development. The incidence and mortality rates of Soviet era have resulted in significant ongoing air and cancer, cardiovascular, and neurodegenerative diseases water pollution, in addition to extensive land contami- is unusually high in Armenia. Out of 183 countries, nation through heavy metal (e.g., lead and arsenic) Armenia has the second-highest cancer mortality rate residue. In addition, wastewater spillages and leakages in the world for both men and women (WHO 2014; have led to contamination of downstream waterways. Bray et al. 2018). The cleantech industry could mitigate some of As the government codifies these health charac- these environmental damages. In addition to alterna- teristics through its digital health record system, tive energy solutions, cleantech solutions include prod- armenia’s software development community can ucts and services that mitigate negative environmental build on the resulting digital assets. Armenia recently impacts. This encompasses a wide range of technolo- became one of the few countries in the world with a gies in reclamation and soil and water treatment nationally centralized digital health record system including biofiltration, sediment capping, soil flushing, (ArMed). ArMed was designed not only to improve dredging, and chemical stabilization. The industry is access to and quality of health care but also to facilitate supported by small and medium innovators and medical education and research. In addition, there has researchers as well as large multinationals to respond been much discussion about the Armenian Genome to topographical, environment and regulatory challenges. Project (AGP) over the past 18 months. The AGP is Not only does the global cleantech industry have 3 significant due to the presumed genetic homogeneity very large growth potential, small and medium of the Armenian population. Armenia’s software enterprises (SMEs) in middle- and low-income development community can integrate AGP and ArMed countries will have critical roles to play over the data to design solutions for global markets as demand next decade. The potential for global cleantech is huge. for digital health care solutions widens to include a Investment across 15 clean technology sectors in 145 variety of users such as physicians, insurance compa- developing countries is expected to top US$6.4 trillion nies, research centers, and pharmacies. over the next decade (World Bank 2014). Critically, 10  |  Strategic GVC Repositioning Options for Armenia over one-third of the total investments (about US$3.59 FIGURE 2.2 trillion) will be directed toward water and wastewater Size of the clean technology market accessible solutions. This market is not reserved for large multi- to SMEs in Eastern Europe national companies only. For SMEs in middle- and — US$ billion low-income countries, US$1.6 trillion of this market is potentially available (see figure 2.1). Wastewater Rather than relying on the local mining industry Small hydro as a market, Armenia can become a concentrated Waste research, development, and prototyping testbed for Water global cleantech firms. Based on current investment in Onshore wind environmental risk solutions, it is unlikely that Solar PV Armenia’s local mining industry can form the corner- Natural gas vehicles stone of a cleantech market. Similarly, Armenia’s public agencies may not yet be in a position to improve or Major equipment enforce environmental regulation in this field. However, Planning installation and balance of system global high-tech firms and research and development Operations and maintenance (R&D) institutions can still utilize Armenia’s factor Source: World Bank 2014. Note: SMEs = small and medium enterprises; PV = photovoltaic. conditions as a concentrated solutions-development and prototyping testbed for cleantech solutions to serve global markets. In Eastern Europe, close to US$30 billion of the clean technology market for wastewater upstream, higher-value-added research and develop- solutions are particularly accessible to SMEs (see figure ment activities, building local capacity and expertise and 2.2). By attracting foreign firms to locate themselves in facilitating innovation in this field. Strengthening Armenia, the government can create a local market for Armenia’s foreign direct investment (FDI) laws and investor “aftercare” would be key to success in this kind of development strategy. FIGURE 2.1 Market size through 2023 for 15 clean EARTHQUAKE ENGINEERING technologies in emerging economies Located on the high activity Alps-Himalaya seismic — US$, trillions zone, Armenia is prone to earthquakes and related hazards. As a result of compression between the Wastewater Arabian and the Eurasian tectonic plates, Armenia is Water home to several geological faults and has a long history Onshore wind of seismic activity. This has included large and destruc- Solar PV tive earthquakes such as the 1988 Spitak earthquake, Small hydro which killed almost 25,000 people, injured 15,000 Waste people, and left 517,000 people homeless. Electric vehicles Mitigating seismic risk has become a high-tech Geothermal global industry. Structural engineering, geotechnical Bioenergy ex feedstock engineering, advanced materials engineering, and Solar CSP nonengineering fields such as catastrophic risk model- Electric bikes Solar thermal ing and insurance markets all form part of a growing Bus rapid transit global industry that targets reduced loss of life and Natural gas vehicles property damage during seismic events. In reflection of Biofuels this, the geotechnical instrumentation and monitoring market is expected to grow to US$5 billion by 2024 (Markets and Markets Research 2019), while the SME Non SME Source: InfoDev 2014. market for seismic reinforcement materials is projected Note: SMEs = small and medium enterprises; O&M = operations and maintenance; PV = photovoltaic. to hit US$40 billion by 2025 (Pulidindi and Chakraborty 2019). 11  |  Strategic GVC Repositioning Options for Armenia Although innovation hubs are concentrated in Service for Seismic Protection, the Institute of high-income economies, middle-income countries Geophysics and Engineering Seismology, and the have leveraged earthquake preparation programs to Institute of Geology) and other nonprofit institutions, develop local capacity. Global innovation hubs are though strong private institutions are starting to concentrated in rich, earthquake-prone economies — emerge (e.g., GEORISK Scientific Research Company). like Italy, Japan, New Zealand, and the United Private sector–led innovations could be further States — where there is a public and private sector catalyzed by targeted research policies including market for risk-mitigation technologies and solutions. applied research grants, competitive research collabo- In addition, other countries like Germany and the U.K. rations, competitions, and R&D tax credits. (neither of which experience high seismic risk) have become global hubs for catastrophic risk modeling due 2.2 to their status as global insurance centers with access to high-performance computational infrastructure. UPGRADING TO “ADJACENT” HIGHER- However, middle- and low-income countries outside of VALUE-ADDED ACTIVITIES these hubs have been able to leverage government earthquake preparation programs to develop local GVC upgrading can come through vertical integra- capacity in seismic risk assessment and structural tion of value-chain activities that are adjacent to retrofitting. Examples include Bulgaria, India, Pakistan, Armenia’s existing ICT services exports. Armenia’s and Romania, which have wide structural retrofitting policy makers can examine the high-tech activities that programs, and Turkey, whose reconstruction program Armenian firms already perform well and implement has also contributed to a local “earthquake engineer- targeted programs to support vertical integration into ing” industry supported by geophysicists, structural “adjacent” higher-value-added activities. This report 4 engineers, geologists, and economists. examines opportunities for upgrading to (a) AI The Armenian government can strategically research and engineering (building on Armenia’s leverage its own earthquake preparation programs competence in mathematics and computer science) and building code regulations to develop Armenia’s and (b) semiconductor design (building on Armenia’s local earthquake engineering capacity. An expansion strong foothold in electronic design automation). of programs like the government’s Seismic Safety Improvement Program (which aims to retrofit and AI PURE RESEARCH build earthquake-resilient schools) could similarly help Because pure AI research (the creation of new to develop geotechnical expertise in Armenia that could algorithms) relies on advanced mathematical serve the Caucasus more widely. Note that stronger research techniques, mathematicians and statisti- building code regulation, inspection and enforcement cians are in high demand in this field. AI algorithms would be key to catalyzing the earthquake engineering can be written in most programming languages, industry. In addition, strengthening these policies including Java, Python and R. As a result, there is a could also support the development of a private sector widespread misconception that pure AI algorithm led insurance market; currently, the information research is a field of computer science or software asymmetry relating to the vulnerability of physical programming. In fact, AI algorithm development uses infrastructure in Armenia has made it difficult for this advanced mathematical research techniques (e.g., important sector to flourish. calculus, linear algebra, probability, statistics, and Targeted R&D investments, policies, and incen- heuristics) that are more common in natural science tives would be key to the development of geotechnical disciplines such as physics, mathematics, and econo- services in Armenia. Beyond earthquake early warning metrics. As the AI market continues to expand, research, Armenia can leverage its location to develop scientists and researchers with deep understanding in new research in fault zone and earthquake geology these fields are in increasingly high demand. fields (e.g., paleoseismology, tectonic geomorphology, Armenia has strong global comparative advan- and seismic imaging) as well as earthquake processes tages in mathematics and natural sciences—even and effects (e.g., crustal deformation, ground motion, more so than in ICT. In 2017, Armenia produced 0.38 and ground failure). In Armenia, this activity is domi- per 1,000 tertiary graduates from ICT-related courses. nated by public institutions (such as the National In the same year, Armenia produced 0.6 per 1,000 12  |  Strategic GVC Repositioning Options for Armenia FIGURE 2.6 NSMS vs. ICT — Number of tertiary graduates per 1,000 people, 2016 /17 Average Algeria India France United States Denmark Canada Puerto Rico Armenia Switzerland Korea Rep Slovenia Germany Georgia Singapore Natural sciences mathematics and statistics Armenia Albania Saudi Arabia Austria Estonia Greece Poland Iran Islamic Rep Tunisia Spain Finland Malaysia Italy Portugal Lithuania Norway Malta Czech Republic Belgium Morocco Sweden Croatia Average Argentina Romania Belarus Russian Federation Kenya Tajikistan Thailand Hungary Bulgaria Serbia South Africa Ukraine Kyrgyz Republic Bangladesh Palestine Kazakhstan Latvia Cyprus Azerbaijan Syrian Mongolia Arab Egypt Ecuador Arab Rep Bosnia and Costa Rica Republic Namibia Oman Herzegovina Mexico Brazil Chile Luxembourg Peru Mauritania Uzbekistan Indonesia Cabo Verde Colombia Uruguay Dominican Republic Madagascar Bahrain Rwanda Bermuda Philippines Burundi Congo Seychelles Dem Rep Qatar Cuba Lao PDR El Salvador Niger Vietnam United Arab Emirates Mozambique Information and communication technology Source: United Nations Educational, Scientific and Cultural Organization’s Institute for Statistics database. Note: NSMS = natural sciences, mathematics, and statistics; ICT = information and communication technology. Armenia 2023 estimation assumes: (a) the distribution of bachelor’s degree programs in 2022 corresponds to the distribution of bachelor’s degree enrollments by program in 2018; (b) the ratio of tertiary to bachelor’s graduates remains consistent; and (c) future tertiary graduate figures are a function of current primary and secondary enrollment figures. 13  |  Strategic GVC Repositioning Options for Armenia 5 graduates in natural sciences, mathematics, and physics. In contrast, AI algorithm researchers need only statistics (NSMS). In both measures, Armenia outper- an internet connection. Where supercomputers are forms global averages (0.32 in ICT and 0.36 in NSMS) needed—for example, for training more complex neural and many of its regional competitors (figure 2.6). networks—they can be accessed remotely. Moreover, between 2016 and 2018, the share of bache- Armenia already has some nascent activity in AI. lor’s degree students graduating from NSMS programs There is already some activity in pure AI research in grew by 50 percent. Tertiary graduates from mathemat- Armenia. Several start-ups support their own in-house ics and statistics programs alone grew by 68 percent. AI development teams for AI-enabled functionality. In 6 Based on current projections, by 2023 9.7 percent of addition, YerevaNN, a nonprofit computer science and Armenia’s bachelor’s degree students will graduate mathematics research lab, recently published its first from NSMS programs. (Armenia’s projected position is peer-reviewed paper in the Nature Scientific Data also shown in figure 2.6.) journal, and another paper is soon to be presented in a Mathematics and natural sciences curricula in workshop organized by the Association for Armenia are current. Industrial technology has Computational Linguistics. On top of this, a number of evolved at a relentless pace over the past 50 years. This Armenian physics professors have published papers dynamism has been particularly challenging in applied relating to AI. science and engineering courses where graduates must However, a lack of post-doctorate training in AI be cognizant of the latest technology breakthroughs, in Armenia is leading to a brain drain and under- conversant in up-to-date industry tools, and display the mining local investment opportunities. The American necessary soft skills and commercial acuity. Globally, University of Armenia offers a bachelor of science mathematics and natural science disciplines have not program in data science, and the Yerevan State 7 advanced at the same pace as applied science and University and the Slavonic University both have engineering courses. As a result, course content for AI-related master’s programs. However, there are no AI these fields in Armenian teaching institutions has been PhD programs in Armenia. Without a PhD program, the able to remain current. industry will struggle to grow as promising researchers Until recently, commercial opportunities for (and firms that draw from this research) look outside mathematicians in Armenia were few. After the Armenia for resources. There is already evidence of collapse of the Soviet Union, many of the institutions this; PicsArt, an Armenian firm (headquartered in the that employed Armenia’s mathematicians and physi- United States) has its own in-house AI development cists disappeared. Job opportunities for team based in Yerevan. However, in October 2018, mathematicians wishing to continue to work in PicsArt opened an AI lab in Moscow with the express theoretical research and related fields are almost purpose of leverage the AI talent pool emerging from entirely confined to universities and teaching institu- the Moscow Institute of Physics and Technology tions where the quality of PhD-level research has been (MIPT). PicsArt intends to enter into R&D partner- called into question. As a result, promising graduates ships with the MIPT to further develop PicsArt’s and researchers have left Yerevan to work in global existing AI-based features, such as visual transforma- research institutions. tion, behavior prediction, and content Unlike in several pure research disciplines, recommendations. competitiveness in AI pure research does not have In addition, corruption has undermined the high capital investment requirements. Traditionally, integrity and quality of Armenia’s PhD programs— the potential for individual researchers and micro even in natural sciences. A 2018 report by the institutions to thrive has been constrained by the need Organisation for Economic Co-operation and for costly laboratory equipment. In the area of applied Development’s Anti-Corruption Network for Eastern physics, for instance, leading universities and research Europe and Central Asia (ACN) discusses the preva- institutions boast cleanroom facilities, microscopy and lence of corruption and corruption risks in the higher reflectometry equipment, processing equipment education system in Armenia. Using official documents, (including deposition, etching, and nanolithography questionnaires, perception surveys, and third-party equipment), and ultra-high vacuum equipment. This is reports by research organizations, the monitoring team not to mention large-scale facilities for research into concluded that corruption is a widespread problem in condensed matter, materials science, and particle Armenian higher education (ACN 2018). Vulnerable 14  |  Strategic GVC Repositioning Options for Armenia areas include university management, human resource FIGURE 2.7 policies, student assessment, licensing, and accredita- Monthly average salaries of faculty in public tion. The assessment also specifically highlights universities in 28 countries, 2012 post-graduate education as an area where gaps in — US$ PPP legislation and institutional structures have facilitated abuse. These risks are acknowledged by the Armenian Canada Italy government, which has specifically targeted corruption South Africa in the Anti-Corruption Strategy 2015–18. (The Anti- India Corruption Strategy 2019–22 is still in draft.) As a result United States of these strategies and consultations between govern- Saudi Arabia ment and civil society, a Programme for United Kingdom Anti-Corruption Measures in Education was agreed Australia upon in early 2018. However, the ACN’s assessment is Netherlands that only three of 12 countermeasures were funded, and Germany that only five of the 12 countermeasures are mandatory. Norway Israel Civil society and research organizations have also Nigeria expressed concerns that the countermeasures are not Malaysia thorough enough. Argentina Low research quality and a lack of replenishment France in the PhD education system will continue to be Japan issues as long as professor salaries remain below the Brazil market rate. Professors, like graduates and research- Colombia ers, can command higher salaries in higher-income Turkey countries. However, this common phenomenon is Czech Republic Mexico exaggerated in the case of university faculty. A 2012 Latvia study by the Center for International Higher Education Kazakhstan at Boston College and the Laboratory for Institutional Ethiopia Analysis at the National Research University Higher China School of Economics in Moscow showed that even Russian Federation when comparing purchasing power parity rather than Armenia real exchange rates, public higher-education faculty members receive very low salaries (Altbach et al. 2012). Entry Average Top Armenians received the lowest compensation of the 28 Source: Altbach et al. 2012. Note: PPP = purchasing power parity. countries in the study (figure 2.7). This issue is compounded by the fact that, whether in Armenia or abroad, faculty can often demand higher salaries in the private sector. As a result of these market failures, already well-developed software programming commu- would-be professors and supervisors are not staying in nity could build a strong offering in AI engineering research, so the PhD system is not replenishing itself. solutions. Already, an Armenian consortium, YEA In addition, with a stronger AI research base, Engineering, is developing a platform for verification Armenia’s software programming and computer and validation of advanced driver-assistance systems, science community could also build a strong offering as well as for end-of-line testing in vehicle-in-the-loop in AI engineering. AI pure research refers to the environments. research and development of new AI algorithms that However, moving from AI research to AI engi- are designed to address hypothetical problems. AI neering is contingent on being given access to engineering refers to the manipulation and adaptation of proprietary data and being able to protect those existing algorithms to “real-world” applications. Data data. In AI pure research, new algorithms are deliber- science and software programming are core compo- ately developed using publicly available data so that nents of AI engineering. With a critical mass of they can be peer reviewed in international fora. researchers and new algorithm development, Armenia’s However, to apply and adapt an AI algorithm to a 15  |  Strategic GVC Repositioning Options for Armenia specific problem (AI engineering), AI teams need Another EDA firm, Mentor Graphics (part of Siemens), real-world client datasets. Few companies in Armenia also entered the Armenian market in 2008 by acquiring have large volumes of digital data (the exceptions the assets of Ponte Solutions Inc. The most recent firm being companies such as GG-Taxi and U-Com). to set up EDA operations in Armenia—Xilinx, Inc.—will Foreign companies will increasingly demand that not be doing so through acquisition. Instead, Xilinx has Armenian firms comply with global information begun a training and employment program to recruit security and management, such as the ISO/IEC software engineers for its Armenia operations. This is 27000-series. Compliance with Europe’s General Data significant because Xilinx will be producing EDA Protection Regulation 2016/679 will also influence the software for field programmable gate arrays (FPGAs), ability of Armenian firms to train neural networks on as opposed to standard chips. FPGAs are dynamically data from private companies within the European programmable, offering higher performance and Union (EU). versatility for a multitude of applications. Xilinx’s The Armenian government can implement decision to locate is operations in Yerevan is testament policies to effectively collect, digitize and share both to Armenia’s core capability in EDA. administrative and statistical data for application Globally, the EDA industry is highly consoli- development. The government may be able to plug the dated—start-ups and small firms are rare. Due to the gap in available data for application development and need for synchronization and standardization across fostering research. Where regular data collection is the sector, the EDA industry is highly consolidated. In already taking place, policies and practices differ 2017, just four companies captured about 68 percent of 9 significantly between ministries, and there is no clear global revenues: Synopsys (valued at US$19.37 billion), 10 policy or necessary infrastructure for efficient data Cadence Design Systems (valued at US$19.26 billion), 11 exchange in the government-to-government, govern- Mentor Graphics (valued at US$4.5 billion), and 12 ment-to-business, and government-to-consumer ANSYS Inc (US$17.58 billion). domains. There is an opportunity for the Armenian Semiconductor chip design—a higher-val- government to standardize the collection and digitiza- ue-added activity—relies on the same core skillset as tion of both administrative and statistical data. The EDA software design but is better suited to start-ups government can also design policies relating to the and small firms. EDA software is used by firms across classification, formatting, storage, and exchange of the semiconductor industry to design chips. Granted these data. More generally, there are no standardized that some of these companies are very large (Intel, mechanisms, policies, or capacities for data-driven ARM, and Samsung being among the largest), but there policy making or encouraging regional cooperation is a greater degree of fragmentation than in the EDA (specifically with the Eurasian Union). To harness industry. “Fabless” semiconductor firms (firms that both internal and international development opportu- design and sell complete chipsets but outsource their nities offered by the global data economy, Armenia fabrication) and intellectual property (IP) core devel- should in parallel develop data policies (including data opers (firms that design IP cores—the building blocks 13 management, cybersecurity, and other policies), of semiconductor chips) are often small- or even amend its legal framework accordingly, and create micro-sized firms. future-proof data infrastructure (broadband networks The potential of start-ups and small firms to and data centers). spin off from Armenia’s EDA firms and compete in semiconductor design may depend on the vertical SEMICONDUCTOR DESIGN application. Three broad applications have emerged Two of the three dominant global electronic design in the semiconductor industry in recent years: (a) automation (EDA) firms have operations in ultra-high-performance, high-volume applications Armenia. EDA software is used by the semiconductor (e.g., consumer electronics, 5G networks and cloud 8 industry to design chips. In 2004, Synopsys entered computing); (b) sensors for IoT proliferation (these do 14 Armenia through its acquisition of Leda Design and not require “latest-node” technology, but it is a Monterrey Arset, both of which were engaged in EDA high-volume, low-margin segment demanding little software design. Synopsys is now Armenia’s largest ICT differentiation); and (c) boutique applications that do employer; with 800 employees, it is the largest of not require high-volume production but instead Synopsys’ offices among its 12 global subsidiaries. demand tailored, customized solutions (e.g., automo- 16  |  Strategic GVC Repositioning Options for Armenia tive). Each segment experiences different competitive venture capital available in Armenia today. Cisco, pressures and therefore exhibits different capital through its acquisition of the U.S. firm Memoir investment requirements, economies of scale, and Systems, is one of the only global vendors with opera- barriers to entry (figure 2.8). tions in Armenia. Although Cisco’s IP design activity Semiconductor and IP design are dominated by represents an important milestone for the country, larger firms in both ultra-high-performance, there is a long way to go to convert Armenia into a true high-volume applications and sensors for IoT research and design hub for semiconductor technolo- proliferation. Faster switching speeds and higher gies. According to Gartner, a research company, a performance are still key drivers in consumer electron- latest-node semiconductor costs US$271 million to ics (e.g., smartphones, tablets, and laptops) and cloud design. Start-ups that are able to raise the necessary computing infrastructure. As a result, the industry has capital rarely see returns for two or even three years, as seen more vertical integration in recent years in the design process is lengthy. Less advanced semicon- applications where high costs can be amortized, and ductors can be cheaper to design, but even 28 large electronic device manufacturers (e.g., Google, nanometer semiconductors (2010 technology) cost Amazon, Huawei, and Panasonic) have started to design US$30 million to design in 2018 (Lapedus 2018). “Soft their own chips for their own specific use. In the IP” only—the initial stage of IP core development— meantime, Samsung Electronics has remained wholly costs about US$10 million, and “Hard IP,” the second vertically integrated throughout the industry’s evolu- stage of IP core development requiring access to silicon tion and performs all functions across the value chain labs for testing and verification, can cost about US$50 from design to manufacture. In the case of IoT, for million (note that sending the first IP core for testing devices to communicate with one another, there must can cost about US$10 million in itself). However, be some degree of homogeneity between them. As a Armenia’s current available investment capital is only result, the industry favors high-volume, undifferenti- about US$15 million, and therefore any gains in ated, low-margin production. semiconductor design would rely heavily on foreign There could be an opportunity for Armenian investment capital or an aggressive ramp-up in venture firms in boutique vertical applications that do not capital funding in Armenia. require high-performance, high-volume production Additionally, Armenian firms that can access but instead demand tailored, niche solutions. funding may struggle to collaborate with leading Semiconductor applications where production at the international firms because the semiconductor latest “process node” is less important and the industry is subject to controls under the Wassenaar emphasis on volume is reduced are more accessible to Arrangement (WA). The WA is an agreement smaller firms in Armenia. Smaller IP core designers between 42 participating states to control exports of and semiconductor designers are today designing all items listed in the List of Dual-Use Goods and bespoke solutions for specific vertical applications (for Technologies and the Munitions List to further interna- example, autonomous vehicles or health) or niche- tional security and stability. The semiconductor enabling technologies. Similarly, there are critical industry receives some of the most scrutiny and limits enabling technologies that are still niche but have exports of not only semiconductor systems, equip- wider industry applications, for example, AI accelera- ment and components but also test, inspection and tors and ultra-low-power solutions. Semiconductor production equipment, materials, and software. and IP design firms offering vertical applications Furthermore, the restrictions go beyond export of differentiate themselves through their understanding tangible goods and include cooperation, collaboration of the client base and downstream hardware and and sharing information. While participating states software technology limitations. In addition, opera- can freely trade in these goods and services among one tional budget and cashflow must be conducive to the another, exports to other countries are controlled product development lifecycle of the target industry. depending on their group designation (A, B, C, D, or 15 Success in this area may depend on the development of E) as defined by the U.S. Department of Commerce. strong and dynamic local (or regional) tech industries Because Armenia is a group D country, WA participat- (see section 2.2). ing states wishing to collaborate and share listed However, venture capital requirements in technologies with Armenian firms need to first obtain semiconductor design are far higher than the export licenses. This obligation increases costs and 17  |  Strategic GVC Repositioning Options for Armenia FIGURE 2.8 Capital expenditure requirements, firm consolidation, and barriers to entry are different in different segments — GROUP 1 HIGH-VOLUME, HIGH-PERFORMANCE MASS-MARKETS Fabless IP core Fabrication EDA semiconductor For example, consumer development plants firms electronics, 5G networks and cloud computing Design services — GROUP 2 SENSORS FOR IOT PROLIFERATION Fabless Undifferentiated design IP core Fabrication EDA semiconductor to facilitate high-volume, development plants firms low-margin production Design services — GROUP 3 A.  NICHE VERTICAL APPLICATIONS For example, autonomous Fabless IP core Fabrication vehicles or health EDA semiconductor development plants firms B. NICHE-ENABLING TECHNOLOGIES For example, AI accelerators, ultra-low- power solutions. Design services High CapEx requiremet, highly consolidated, high barriers to entry Medium CapEx requirement, medium consolidation, medium barriers to entry Lower CapEx requirement, lower consolidation, lower barriers to entry Source: Author’s own elaboration. Note: AI = artificial intelligence; EDA = electronic design automation; IoT = internet of things; IP = intellectual property. 18  |  Strategic GVC Repositioning Options for Armenia reduces the ease with which the global semiconductor number of technology giants (such as IBM, Google, and industry can collaborate with Armenian firms in these Intel) and start-ups backed by venture capital (such as cutting-edge technologies. Because EDA is not covered Rigetti, IonQ, and Quantum Circuits) have been able to on the lists, this has not been a problem so far. But as build prototype quantum computers or hardware Armenia seeks to upgrade participation in semicon- subsystems. Although some universities and research ductor value chains, the Armenian government may institutions may have also built their own prototypes, consider working with the U.S. Department of given the complexity and capital requirements involved Commerce to address security concerns and improve in recreating quantum conditions (let alone in a way Armenia’s grouping. that is “programmable”), quantum computing proto- type construction will likely remain highly consolidated 2.3 for the time being. However, outside of hardware development, the BUILDING CAPACITY AND RELEVANCE IN quantum research ecosystem is vast, and the global NASCENT TECHNOLOGIES WHILE THEY playing field is level. Many of the most prominent ARE STILL ACCESSIBLE firms that have constructed prototype quantum computers also offer free access to their prototypes via QUANTUM COMPUTING R&D the cloud. IBM, Rigetti, Google, Alibaba, Microsoft, and Quantum computing research is still in its infancy. D-Wave all offer such services, allowing users around In the early 1900s, a number of theories were put for- the world to run algorithms and experiments, and to ward to explain peculiar observations in the behavior of publish research derived through each platform. In so subatomic particles. Up until that time, the assumption doing, hardware developers are able to “crowdsource” was that two-state systems, such as the spin of an elec- development and at the same time build an open tron (up or down) or the polarization of a photon (hor- source community around their solutions as the race izontal or vertical), could only exist in one state or the toward quantum supremacy continues. Meanwhile, the other. However, it was discovered that, at atomic scales cloud computing model has meant that physicists, data and smaller, some systems (called qubits or quantum scientists, and mathematicians have been able to begin bits) can exist in both states at the same time. These now- to research algorithms for a range of vertical applica- proven theories form the basis of quantum computing; tions. There are also opportunities in “upstream” a computer that could harness these (and other) sub- development, for instance, in core infrastructure atomic behaviors would be exponentially faster than redesign, performance enhancement or documentation existing “classical” computers. The possible implica- improvement. As an example, one of the limitations of tions for the global economy are vast, with applications quantum computers is their susceptibility to changes in across the economic spectrum, including finance, energy, their surrounding environment and their consequent pharmaceutical, and logistics. The challenge has been tendency to produce errors. Small start-ups and in recreating these behaviors in the physical world in a independent researchers are actively engaged in way that can be controlled and manipulated. There are designing software for error mitigation and are attract- many possible approaches, but over the past 20 years ing significant investment in doing so. research has coalesced around five or six. Several pro- State-sponsored quantum programs historically totype quantum computers have now been constructed formed the backbone of quantum computing 16 based on these approaches, but at the time of writing, research. Just as quantum computing promises a range no prototype has been able to overcome fundamental of economic benefits, quantum computing also noise issues, making quantum computers no more pow- threatens the encryption systems that are ubiquitous in 18 erful than classical computers. The global quantum today’s digital infrastructure. This includes the global research community is therefore yet to demonstrate finance and monetary systems, digital health systems, “quantum supremacy”—proof that quantum comput- and other systems that depend on data protection and ing devices can solve problems that would otherwise be secrecy. These concerns have prompted governments 17 impossible for classical computers to tackle. to invest heavily in quantum computing. China leads Only a handful of entities globally have had the the way, having announced plans to open a 92-acre, capital, knowledge, and convening power to US$10 billion quantum research center by 2020. In construct prototype quantum computers. A small direct recognition of the data security concerns, the 19  |  Strategic GVC Repositioning Options for Armenia United States passed the US$1.275 billion National automatic quantum compilation and programming. Quantum Initiative Act in 2017. The EU similarly Conceptually, both activities seek to map a theoretical launched the Quantum Technologies Flagship in 2018, a computational algorithm to a physical computing 10-year €1 billion research initiative. Other countries device. Some of the most prominent approaches to with large state-sponsored programs include Australia, quantum computing—used by IBM, Google, and Canada, Israel, Japan, Sweden, and the United Kingdom. Intel—use semiconductors as an integration platform However, private sector funding for quantum for their fundamental information building blocks research now dominates, opening the door to (qubits). If Armenia is able to build on its core compe- researchers in middle- and low-income countries. tence in EDA and move toward semiconductor design Large multinationals and venture capital funds have itself, there may also be opportunities for crossover and dwarfed state-sponsored quantum research in recent collaborative activities and research. years. IBM invested US$38 billion in a quantum Nonetheless, Armenia has no national quantum computing program between 2013 and 2018 (Green strategy and is not leveraging its potential compara- 2018). But research into quantum computing has also tive advantage in this area. There is only one revealed commercially valuable discoveries in other quantum computing lab in Armenia at the time of fields (e.g., quantum communication, quantum metrol- writing this report, and no formal training at the ogy, and quantum simulation). Private firms that may bachelor’s, master’s, or doctoral levels. As the global not have been actively engaged in developing their own race intensifies, Armenia may need to develop a quantum solutions are now funding independent quantum computing strategy and road map, at the very research. As a result, research labs in middle- and least for reasons of security, and to stake its position. In low-income countries that are unable to benefit from so doing, the government could consider policies that state-sponsored opportunities have much greater encourage the international research collaboration in prospects of participating in global quantum computing the field of quantum research, position Armenia development, armed with little more than a laptop and globally as a regular “stop” on the quantum computing an internet connection. conference circuit, introduce student sponsorship and Global citations of Armenia’s R&D in physics far exchange programs and, where possible, make targeted exceed those for mathematics and computer science, investments in quantum computing research. putting Armenia in the top 25 countries in the world for R&D citations in nuclear physics. As described previously (see section 2.2), Armenia’s teaching institutions have struggled to keep pace with dynamic disciplines such as engineering and computer science. However, global advances in natural sciences, particu- larly foundational physics and mathematics, have been slow, or even stagnant for 50 years. Despite concerns about integrity in Armenia’s tertiary education programs, Armenia’s physics curricula and research output remain industry-relevant. This includes PhD-level research and is evidenced by the number of (internationally) cited documents that Armenia’s physics community publishes each year (figure 2.9). In 2018, Armenia ranked 25th in the world for the number of citations related to nuclear and high energy phys- 19 ics—key facets of quantum physics. Armenia’s strong background in EDA is also highly relevant to quantum computing development. EDA software is used to design semiconductors, and Armenia already has a strong foothold in this area (see section 2.2). Several start-ups and research institutions are investigating the use of EDA techniques for fully 20  |  Strategic GVC Repositioning Options for Armenia FIGURE 2.9 Citations of Armenian research papers in international journals, 2018 Medicine Multidisciplinary Cites per document Physics and astronomy Engineering Environmental science Mathematics Chemistry Neuroscience Documents Source: Scimago Institutions Rankings database. 21  |  Strategic GVC Repositioning Options for Armenia Case study ICT SERVICES FOR DIGITAL HEALTH AND BIOTECH APPLICATIONS of 38.4 percent across Europe and a Armenia’s potential to Central Asia. deliver ICT services for biotech applications should However, the global market for not be constrained by the digital health applications is lack of local demand projected to reach US$504.4 billion by 2025. Armenian start-ups are The local market for digital health already providing some hospital solutions is underdeveloped in management solutions to foreign Armenia. Government per capita clients, but globally, digital health spending on health care is compar- includes a vast array of markets atively low (table CS2.1). Though and applications. Consumer-facing private spending is quite high, the mobile applications have enjoyed a preventative digital health care solu- surge in popularity in recent years, tions market is still in its infancy in focusing on wellness management Armenia. As a proxy for the reduced considerations such as diet, nutri- market consumer awareness and tion, lifestyle, and fitness. Other concern for wellness, consider that “health condition management” 52 per­cent of Armenian men smoked solutions are more clinical (for in 2016 compared with an average example, medication management TABLE CS2.1 Health care expenditure per capita, 2016 — Current US$ PPP Private Public Armenia 294 59 Europe and Central Asia 467 1,625 Central Europe and the Baltics 239 613 Small states 190 376 World 260 763 Source: World Development Indicators database. Note: PPP = purchasing power parity. 22  |  Strategic GVC Repositioning Options for Armenia or women’s health and pregnancy different health care providers physicians, insurance companies, solutions). However, digital health operate different systems, access- research centers and pharmacies. development includes many more ing the data can be bureaucratic Moreover, as consumer demand technologically complex applica- and complex. Even countries such for health applications grows, tions in areas in which Armenia is as the United Kingdom, which has Armenia’s software development not yet participating. These areas a national health care authority, can community can utilize EHR data include diagnostics and treatment, have problems; as of April 2017, a to research and develop mobile disease and epidemic prediction case study by the Centre for Public health and wellness applications. and tracking, and remote moni- Impact found that many hospitals Solutions to protect data and the toring devices and services. In in the U.K. still lack a compre- privacy of health care records are 2018, digital health firms in the hensive electronic patient record another a growing subset of the United States alone attracted system (Basu 2017), and different digital health sector on which US$8.1 billion of investment — a health care trusts may be served by Armenian software developers may 42 percent increase over the different information technology also be able to capitalize. previous year (Day and Zweig (IT) suppliers. ArMed is a single 2019)—and the global market could nationwide system, accessible reach US$504.4 billion by 2025 through one portal, collating data (Ugalmugale and Swain 2019). from 474 organizations, including Armenia’s population has hospitals, dental clinics, diagnostic distinctive health centers, and insurance companies. characteristics that could be valuable to global biotech Digital health development In addition, ArMed has been delib- development is reliant on access to data erately designed to make its data and strong software conducive to digital health applica- A global rise in chronic diseases development capacity— tion development. When sourcing is fueling growth in the market Armenia has both data from several different EHR for remote patient monitoring systems, issues relating to compat- solutions. Aging populations, In December 2017, Armenia began ibility or comparability can arise. urbanization, and the shift from nationwide implementation of the This is especially true when data- labor economies to knowledge zArmenia National Electronic bases are designed with differing economies has led to a rise in the Health Information System emphases (e.g., patient engagement incidence of chronic illnesses such (ArMed). In 2012, the government vs. performance measurement vs. as diabetes, cardiovascular diseases, designed and approved a program reducing costs). In contrast, ArMed and asthma. This increased prev- to introduce an e-health system. was designed not only to improve alence is fueling demand for After five years of development access and quality of health care but remote patient monitoring solu- and a successful pilot in six health also to facilitate medical education tions. Remote patient monitoring care facilities, a National Electronic and research. allows patients and clinicians to Health Operator was established in track medical and health-related December 2017, and the program Start-ups in Armenia’s software information. Estimates differ, but, began rollout. As of June 2019, the development community can use on average, the global market for system is already serving 55 percent ArMed data to design solutions for remote patient monitoring tech- of the population. global markets. Armenia has strong nologies is expected to rise to about b competence in software develop- US$2 billion by 2023. Unlike many other digitized health ment and is gradually moving away records, ArMed data are acces- from an outsourcing model toward Chronic diseases are unusu- sible through one centralized original products and solutions. ally prevalent in Armenia. Sadly, database. Several countries have Growing adoption of health care Armenia is an outlier with respect electronic health records (EHR) IT solutions globally will increase to the incidence, prevalence, and systems. However, in countries demand for software solutions mortality rates of chronic nonin- such as the United States, where for a variety of users, including fectious diseases such as cancer, 23  |  Strategic GVC Repositioning Options for Armenia cardiovascular illnesses, and neuro- Armenian patients could participate on the growing chronic health degenerative diseases. Out of 183 in clinical trials, provide feedback, community for their experience countries monitored by the World and influence the design of these and insight, as well as for clinical Health Organization, Armenia has devices globally. data. the second-highest cancer mortal- ity rate in the world for both men Data related to chronic diseases and women. In men, the age-stan- can also inform the design of dardized rate reached 354.5 deaths preventative health care and tele- The presumed heterogeneity per 100,000 persons in 2016 (figure health software and hardware of the Armenian genome c CS2.1). This is compared with a solutions. New business models could give Armenia a 2018 average of 83.4 deaths per such as preventative health care comparative advantage in 100,000 persons in countries with and telemedicine are starting to the research and a medium Human Development gain traction at both ends of the development of further Index, and a world average of 122.7 economic development spectrum applied bioscience solutions deaths per 100,000 persons (Bray through their potential to reduce et al. 2018). health care costs. Preventative Genome sequencing has become health care business models place a core area of applied biosciences. The design and development of greater emphasis on preventative Genome sequencing refers to the remote patient monitoring biotech care and wellness than traditional large-scale comparison of DNA, is reliant on access to patients with health care business models. In sequences of populations. The chronic diseases, in addition to raw preventative health care, care plans scale and complexity of genome data. Remote patient monitoring are personalized to the individual, sequencing projects have meant technologies and solutions include and personalized treatments are that many have been publicly implantable devices (e.g., cardio- delivered at a precise time and funded. National genome sequenc- verter defibrillators), wearable place, as appropriate to the patient ing projects enable scientists to devices (e.g., continuous glucose in question. Telehealth solutions characterize the population-level monitors and cardiac monitors), go beyond remote monitoring genetic background and inves- and internet of things solutions devices to include (for instance) tigate the prevalence of various such as sensors for smart watches communication technologies for traits and disorders (Arakelyan and mobile applications. Patients “real-time” telemedicine (including and Glorikian 2019). This area with chronic diseases are crucial video and audio medical assess- of genomics has applications in stakeholders in the research and ments and basic examinations), digital health, biotechnology, envi- development of remote patient as well as diagnostic imaging and ronmental applications, forensics, monitoring biotech solutions. In interpretation solutions. Both and security (Batelle Memorial addition to the concentrated pool preventative health care solutions Institute 2013). Currently, 50 of data that ArMed is generating, and telehealth technology depend countries around the world have population sequencing programs. FIGURE CS2.1 Male cancer mortality rates in 183 countries, 2016 The Armenian genome may be distinct in that Armenia’s population is thought to be genet- Armenia United ically homogenous. The Armenian India States Georgia China government has stated its intention to embark on an Armenian Genome Project (AGP). This could be significant as there is some expec- tation in the genomic research community that Armenia has a Mortality rates per population genetically homogenous popula- Source: World Health Organization’s Global Health Observatory data repository. tion. Homogenous genomes are 24  |  Strategic GVC Repositioning Options for Armenia easier to use as a basis for research, A complementary FDI strategy as heterogeneity can create “noise” With an appropriate could further attract global industry making results more difficult to regulatory framework, stakeholders, making a significant isolate and interactions easier to Armenia can position itself contribution to local capacity devel- identify. as a global hub for digital opment. A targeted investment health, biotech technology promotion and attraction strategy If the data from the AGP were to be development and bioscience could complement the policies integrated with Armenia’s ArMed research described herein. World-class platform, the potential could be research and medical institutions, even greater. Integrating data The Armenian government would pharmaceutical companies, and from the AGP could amplify the need to carefully draft localiza- biotech organizations could make potential of both initiatives. Not tion and data sovereignty laws that significant contributions to local least, it would enable the compar- define limitations for accessing capacity development (figure ison of individual patient variants data from outside Armenia. Many of CS2.3). with AGP variants and facilitate the opportunities described in this — outcome monitoring at both the case study are predicated on the a. World Development Indicators individual and aggregate level. This database. assumption that firms must locate b. Coherent Market Insights (2017); could help identify novel genetic themselves in the country in order Sumant and Jaiswal (2016); Grand View variants linked to chronic diseases to access data. However, rather Research (2019). prevalent in Armenia, such as than placing a blanket ban on data c. Jobs number does not include 2012 cancer, cardiovascular disease and access, the Armenian government figures. neurodegeneration (Arakelyan and has the opportunity to draft local- d. World Health Organization’s Global Glorikian 2019). Health Observatory data repository, ization and data sovereignty laws http://apps.who.int/gho/data/view.main. that do not stifle interest but still A1110. The economic impacts of human stimulate local economic activity in genome projects in other coun- Armenia. tries have been significant. Battelle Memorial Institute (2013) has The Armenian government and the estimated that, between 1988 and administrators of ArMed and AGP 2012, the U.S. federal government’s might consider designing reach- investment of US$3.8 billion in the through licensing agreements and Human Genome Project generated other smart solutions to monetize economic output of US$965 billion, the data. “Reach-through” agree- US$462 billion of value added, ments allow upstream technology personal income exceeding US$293 firms to earn royalties from sales billion, and more than 310,000 or use of downstream solutions d jobs. San Diego, a city of 1.4 based on their technologies. The million people, is the most patent practice is most common in phar- intensive genomics ecosystem in maceuticals and could have wider the United States and received application as the discipline merges nearly one-quarter of all genom- with other sectors as part of the ics venture capital in the United digital and preventative health States in 2016. An input-output movements. The government and study by BW Research calculated the administrators of ArMed and the annual economic impact of its AGP may consider this and other genomics industry to be US$5.6 innovative monetization strategies billion, supported by 115 core to provide firms with incentives genomics firms employing over to pay to access Armenia’s data 10,000 people (San Diego Regional resources. EDC 2016). 25  |  Strategic GVC Repositioning Options for Armenia FIGURE CS2.2 Structure of the digital health GVC (at intersection with biotech and genomics) GLOBAL DNA Pharma R&D Wearables Implantables ACTIVITIES sequencing and drugs and software/ software/ genomics diagnostics hardware design hardware design Bioscience Regenerative Bioinformatics Signal Manufacturing VC and Integrated R&D/lab medicine processing and services early stage system services sensor design financing architecture Data science B B software Diagnostic and B C software research and development imaging development KEY services services Value chain activity Relative factor Intensity factor of intensity production Labor Low Capital Medium Knowledge High Legal and Integrated case Medical lab Acute/ Nursing/ Behavioral insurance management test services primary care palliative care/ health coaching Value chain links Arms length Frequent services services physical therapy and incentive ACTIVITIES trade communication/ management CLOSER TO LOCAL relationships collaboration Time critical delivery HEALTH SERVICES hours/days MARKET Services/products to order weeks/months PUBLIC PUBLIC DISEASE BUNDLED GOODS CERTIFICATION SKILLS INCENTIVES PUBLIC HEALTH PUBLIC R&D HEALTH REGISTRY OUTCOME AND AND AND TAXATION COMMUNICATION epidemiology POLICY MONITORING BASED ACCREDITATION EDUCATION REGIMES outcomes PAYMENT Source: Author’s own elaboration. Note: B2B = business-to-business; B2C = business-to-consumer; R&D = research and development; VC = venture capital. FIGURE CS2.3 Armenia’s participation in the biotech/digital health value chain GLOBAL DNA Pharma R&D Wearables Implantables ACTIVITIES sequencing and drugs and software/ software/ genomics diagnostics hardware design hardware design Bioscience Regenerative Bioinformatics Signal Manufacturing VC and Integrated R&D/lab medicine processing and services early stage system services sensor design financing architecture Data science B B software Diagnostic and B C software research and development imaging development KEY services services Value chain activity Relative factor Intensity factor of intensity production Labor Low Capital Medium Knowledge High Legal and Integrated case Medical lab Acute/ Nursing/ Behavioral insurance management test services primary care palliative care/ health coaching Value chain links Arms length Frequent services services physical therapy and incentive ACTIVITIES trade communication/ management CLOSER TO LOCAL relationships collaboration Time critical delivery HEALTH SERVICES hours/days MARKET Services/products to order weeks/months PUBLIC PUBLIC DISEASE BUNDLED GOODS CERTIFICATION SKILLS INCENTIVES PUBLIC HEALTH PUBLIC R&D HEALTH REGISTRY OUTCOME AND AND AND TAXATION COMMUNICATION epidemiology POLICY MONITORING BASED ACCREDITATION EDUCATION REGIMES outcomes PAYMENT Source: Author’s own elaboration. Note: The figure shows Armenia’s participation in the biotech/digital health value chain. Armenia is already participating in downstream software development activities (highlighted in yellow) but is yet to participate in higher-value-added upstream activities including bioscience R&D and lab services, bioinformatics, wearables or signal processing. Close-to-market, higher-value-added activities like “integrated case management” require presence in target market and local expertise. Chapter 3 STRENGTHENING THE ECOSYSTEM A stronger regulatory environment is key if ment several financing programs and grant schemes. Armenian firms are to successfully compete for Since then, other funds have emerged to support higher-value-added activities. Structural weaknesses Armenia’s start-up ecosystem. Armenia’s first venture in Armenia’s regulatory ecosystem are constraining the capital fund, Granatus Ventures, was launched in 2013, potential for start-ups and young firms to participate and the second one, HIVE Ventures, was launched in more in high-tech global value chains (GVCs). First and 2014. In more recent years, several other schemes and foremost, funding gaps and legal restrictions have made platforms have emerged, including SmartGate Venture 20 local incorporation unattractive. At the same time, leg- Capital. Although new funds are in discussion, islation governing foreign direct investment (FDI) is Armenia’s venture capital envelope is estimated at somewhat outdated, limiting inward investment. As a US$15 million (as of July 2019). With little publicly result, Armenian firms in high-tech sectors have incen- available information and sustained inconsistencies in tives to incorporate abroad even when their operations the reported value of each of these funds (during are to remain in Armenia, limiting local tax revenue and presentations, events, consultations, and interviews), spillovers. Further upstream, Armenia’s educational their value may be much lower. institutions are not adequately preparing students. At the same time, FDI opportunities are These themes are discussed in the following sections. constrained by a 1994 law that does not adequately address current investment realities and limits 3.1 arbitration to Armenian courts. Protection of foreign investments is governed by the law on Foreign INNOVATION AND ENTREPRENEURSHIP Investments of 1994 (or on a country-by-country basis through bilateral treaties). Drafted 25 years ago, the law There is only limited venture capital available in does not adequately address the current international Armenia to support business innovation and investment landscape. A revised law could offer more entrepreneurial endeavors. Armenia has recently predictable, accessible, and effective protection experienced a wave of early stage financing schemes. It mechanisms for investors (e.g., defining preconditions started in 2006 with the Enterprise Incubator for resorting to expropriation or defining characteris- Foundation’s (EIF) Science and Technology tics of compensation payment). Investor “aftercare” Entrepreneurship Program and it continue to imple- could be improved by introducing clear mechanisms for 28 management of grievances and complaints. Crucially, capital (World Bank 2019). As a result, several initia- changes to the law could allow investors to use interna- tives for angel and seed-stage funding have started, tional foreign courts rather than Armenian courts in including the Business Angel Network of Armenia, the the case of arbitration, if they prefer. Fast Foundation’s Science and Technology Angel As a result of these limitations, Armenian firms Network, and the Angel Investor Club of Armenia, often incorporate abroad to attract investment, among others. The aforementioned EIF is one of the limiting local spillover effects. The lack of early stage largest technology business incubators in Armenia, financing in Armenia has driven several Armenian tech which aims to create a productive environment for firms to seek investment abroad. In most cases, innovation, start-up creation, and higher productivity. investors explicitly request investees to incorporate In addition, several other incubators, accelerators, under laws that the investors are familiar with. This is coworking spaces, business centers, and entrepreneur especially true of U.S. investors. Given that many associations support start-up growth in Armenia. Armenian firms have grown through U.S. investments However, thus far, the measurement of success for (due in large part to strong diaspora links), most many of these schemes focuses on the number of start- Armenian tech start-ups are incorporated in the United ups and entrepreneurs that participate in the scheme States, even if their U.S. headquarters are virtual or rather than follow-on funding, sales, or business viabil- only nominally staffed. This firm structure reduces the ity (of the scheme). As a result, there is very little idea Armenian entity to subsidiary status, meaning that any validation across the ecosystem. In Silicon Valley in the investment they receive remains in the U.S. account United States and other global tech hubs, start-ups with and U.S. dollar transfers are limited to operational poor market viability would not usually be in an incuba- costs, salaries, and overheads. As an example, PicsArt tor for more than four to six months, because the incu- (valued at an estimated US$250 million) was able to bator framework recognizes that the start-up is nonvia- raise capital through two rounds of venture capital ble and parts ways with entrepreneurs. Conversely, funding by incorporating in the United States. industry stakeholders in Armenia have repeatedly Shortcomings in Armenia’s deal flow manage- raised concerns during consultations that nonviable ment and seed-funding schemes are also constrain- start-ups circulate through the ecosystem’s incubator ing Armenia’s high-tech start-up growth. Many and accelerator programs over and over, compromising Armenian start-ups are too small to attract venture the efficacy of the ecosystem. As the Armenian govern- 29  |  Strengthening the Ecosystem ment, together with the development community, con- financing from commercial banks and only 15 percent tinues to expand these programs, more emphasis can sought funds from credit organizations. Put differently, be placed on managers, trainers, and mentors (in addi- almost three-quarters of firms were financing their tion to the scheme’s monitoring and evaluation struc- investments internally in fixed assets and working ture) to bolster the efficacy of the programs. capital rather than through banks. Some aspects of Armenia’s corporate law are Promotion of financial technology innovations poorly suited to early stage financing and invest- (e.g., mobile banking) and use of agents could help ment. In reflection of their added risk, venture capital expand outreach to underserved segments and and angel investors often expect to acquire preferred locations. The government can develop a fintech stock (preference shares) with liquidation and dividend strategy to bridge financing gaps for small firms and preference. Early stage investors may also demand expand outreach at reduced costs. In doing so, leading some voting power (even if subordinate). Armenian law finance institutions can better address information only allows joint-stock companies to issue preferential asymmetries and reduce risks in lending, facilitate stock and must be “nonvoting” shares. There is no legal market monitoring, and introduce innovative products. structure that permits different classes of voting shares, With adequate regulatory safeguards and oversight, which entail differing entitlements to dividends and agents (e.g., the Armenian Post) could support delivery different levels of voting power (IP/IT Committee for of core financial services through partnership or agency Armenia 2019). Additionally, where seed funding is agreements with financial institutions. received as a debt, converting that investment into At the same time, small firms need additional equity during subsequent funding rounds (as is com- business and financial management training to take mon with convertible loans) is not possible under advantage of such schemes. Low levels of product Armenian law. awareness and financial management capacity in the Finally, intellectual property is not well protected. MSME sector are considerably affecting firms’ ability to Attitudes to patenting software differ from industry to select adequate financial instruments. Additionally, industry. Many industries rely on interoperability and business, financial, and strategic planning are rare. In open source code, making patenting redundant. Other general, investment plans are frequently developed industries rely on alternative areas of their business to without a broader long-term vision for the enterprise or create and retain value (e.g., brand recognition, a solid assessment of the viability, risk, or expected language and culture, service delivery). However, cash flow of the investment. As a result, financial outside of mass-market applications (e.g., mobile apps, institutions report that smaller enterprises often apply games, health management software, etc.) software for loans they cannot afford and face significant patenting is more common. A software that is unassoci- challenges in repayment. ated with a hardware-based application has traditionally There is already a high level of informality in the been difficult to defend in a court of law, but a software enterprise sector, which may be exacerbated in 21 showing “further technical effect” (i.e., one that is software development activities. A 2018 International integrated with hardware) is more defensible. Armenia Monetary Fund study (Medina and Schneider 2018) has neither legislation to protect “standalone” software estimates Armenia’s informal economy at around 36 nor software that is integrated with hardware. percent of its overall gross domestic product (GDP) in 2015, slowly declining below 40 percent since 2011. 3.2 Although Armenia’s shadow (i.e., informal) economy is much smaller than other countries in the region ACCESS TO GROWTH FINANCE (Georgia’s is 53 percent while Ukraine’s is 43 percent), it is significantly larger than in other comparable Small firms in Armenia rely heavily on internal countries such as Bosnia and Herzegovina (30 percent), resources and retained earnings for growth finance. Albania (26 percent), or Croatia (23 percent) and even In 2013, the main source of financing for micro, small, double the size of those in Estonia and the Islamic and medium enterprises (MSMEs) in Armenia were Republic of Iran (18 percent). Software development the owner’s own funds and company resources (World activities, where barriers to entry are low, are suscepti- Bank 2013). Friends and relatives were other major ble to informality. In addition, cash-on-delivery is the sources of financing. Only 55 percent of firms reported most popular payment method in the nascent e-com- 30  |  Strengthening the Ecosystem merce market. (As an example, only 9 percent of credit Intermediary liability regulations distribute liability card payments are attributed to online transactions). between intermediaries (e.g., websites and apps) and Pervasive informality will affect the potential of firms online vendors or content developers in the event of a to access to credit, because lenders will not be able to dispute. Rules on intermediary liability need to strike a use cash flow-based credit analysis to assess viability. balance between protecting consumer rights and supporting the expansion of digital markets, including 3.3 through intermediary platforms. Digital intermediaries manage the relationship with the customer, and they are DIGITAL TRADE often the largest, most sophisticated actors involved in a transaction. As such, regulations can impose intermedi- Armenian law recognizes e-documents and digital ary liability for fake or faulty products or services, or signatures but could go further to allow contracting for offensive or illegal content transacted through their parties to select the technology that best suits their services. Armenia lacks a specific framework to address needs. A conducive regulatory framework for digital online intermediary liability. The government could trade should guarantee that contracts concluded consider a framework that explicitly addresses online remotely through electronic channels are valid and intermediaries, with safe harbor provisions to shield legally enforceable just as those concluded in person. them from liability for infringing content posted by Armenia’s Law on Electronic Document and Electronic third parties, including intellectual property violations. Signature and the Civil Code recognize electronic documents as equivalent to paper-based documents. 3.4 Additionally, the law is technology-neutral regarding storage of e-documents (allowing parties to decide RESEARCH AND EDUCATION what is most appropriate based on their circumstances). 22 However, the law only recognizes digital signatures as The Armenian government’s support for research legally valid rather than allowing contracting parties to and development (R&D) has largely focused on select the technology that best suits their needs. The infrastructure investments. Public funding for Armenian government could consider expanding the Armenia’s science, technology, and research potential law to recognize all electronic signatures and could, in have thus far been focused on infrastructure (e.g., addition, grant them “presumption of validity and technology parks, centers of excellence, labs). authenticity” in legal proceedings. Examples include the Gyumri Technology Centre and Cybersecurity provision could enhance the Vanadzor Technology Centre. More recently, Armenia’s already strong framework for online “Engineering City” in Yerevan is under construction, data protection. In 2015, Armenia introduced a aiming to create an environment for engineering firms comprehensive legislation to protect online personal to accelerate the development of complex engineering data privacy, taking elements from the European Union solutions. Facilities will include research and proto- Data Protection Directive of 1995. The Law on Personal typing labs, machine tooling, and advanced Data Protection includes important provisions such as production equipment. special rules for the treatment of sensitive personal There is an opportunity to prioritize research data, legal bases of data collection and processing, data areas and implement targeted R&D policies. Sections subjects’ rights to access and deletion, and rules on 2.2 and 2.3 detail Armenia’s potential in pure and cross-border data transfers. However, although the law applied research in fields related to natural sciences includes a framework for cybersecurity with require- and mathematics. Notably, Armenia has high potential ments such as data encryption and breach notification, in the areas of AI pure research and quantum comput- it lacks certain recommended provisions. The Agency ing R&D. Armenia’s potential in high-tech sectors is for the Protection of Personal Data could consider closely tied with its R&D framework. An increase in including specific regulations for cybersecurity, adding public funding for R&D would also be necessary to requirements such as the appointment of a data realize Armenia’s potential; the country spent 0.2 manager and risk assessment procedures. percent of its GDP on R&D in 2017 compared with a The Armenian government does not yet have a world average of 2.2 percent. Additional budget could framework that addresses intermediary liability. fund targeted research policies including applied 31  |  Strengthening the Ecosystem research grants, competitive research collaborations, Systems Engineering; the establishment of the competitions and R&D tax credits. There are plans to Armenian National Engineering Laboratories at the develop 24 separate sectoral strategies including a NPUA which upgraded outdated laboratory equipment High-Tech Industry Strategy (under the Ministry of and modernized the engineering curriculum; and the High-Tech Sectors) that will be separate from the Armenian-Indian Center for Excellence in Information Education, Science and Research Strategy (under the and Communication Technologies established in Ministry of Education, Science, Culture and Sports). Yerevan state University. However, as the World The Armenian government should go beyond develop- Development Report 2020: Trading for Development in the ing “complementary” strategies and rather intertwine Age of Global Value Chains notes, in lieu of public sector the two strategies. progress, employer-sponsored training within GVCs Armenia’s engineering and computer science can become a mechanism for skills development. This curricula need improvement and regular review. is clearly evident in the case of Synopsys, which Armenia’s engineering and computer science courses operates its own bachelor’s, master’s, and research are severely lagging industry. Engineering courses are programs at its educational center and at the NPUA, particularly out of date, with few courses incorporating Yerevan State University, the Russian-Armenian computer-aided design software (e.g., Autocad or (Slavonic) University, the European Regional Academy Solidworks) or printed circuit board software (e.g., and the National Research University of Electronic Altium or Orcad). Although computer science disci- Technology. Not only are Armenia’s engineering and plines have seen some improvement in recent years, computer science courses in need of overhaul, public graduates are still not industry-ready or even “indus- universities can better work with the private sector to try-relevant” in some cases. As an example, prominent provide more professional development and on-the-job universities such as the National Polytechnic University training options that provide mid-career learning of Armenia (NPUA) are still teaching Turbo Pascal, a options. Providing training through nimble and programming language developed in the 1960s that was dynamic learning tools—such as certificate programs, superseded by C, C++, and Java in the 1990s. Individual single classes, and massive open online courses—will universities are also taking the lead themselves; e.g., help universities to adapt quickly to changing and NPUA’s partnership with Rutgers University, National emerging technologies. Instruments and EIF to offer a master’s program in 32  |  Strengthening the Ecosystem Chapter 4 ACTIONS AND RECOMMENDATIONS This chapter introduces recom- to employ change management mendations that could strengthen approaches to maximize the Armenia’s participation in high- impact of reforms. As an example, tech value chains through policy public-private dialogue mecha- reforms, capital investments, and nisms could raise awareness and skills development (table 4.1). understanding of the reforms and As the discussion in the their implications. The govern- previous chapters has demon- ment must also make sure to strated, any strategic activity to allocate adequate resources to reposition Armenia in high-tech enforcing changes in the law where value chains will require strong appropriate (e.g., changes in administrative and coordination corporate law) so as not to capacity in government. In undermine the reform effort. particular, the Ministry of High- Tech Industries and the Ministry of Education, Science, Culture and Sports need a joint operating framework that will allow them to codesign and co-implement an overarching innovation strategy for Armenia. It is worth noting also that a simple change in government policy may not be enough; the Armenian government may need 33 TABLE 4.1 Summary of recommendations RECOMMENDATION AUTHORITY TIMEFRAME Entrepreneurship, research, and innovation » Develop an innovation policy road map that identifies policies MoESCS, MoHTI Long term that encourage international research collaboration in priority fields. » Provide support for research and innovation programs through MoESCS, Short term subfinancing, e.g., applied research grants, competitive research MoHTI collaborations and matching grants programs for start-ups. » Strengthen human resources, research excellence, and research MoESCS Short term commercialization through government-facilitated collaborative research programs with international universities, private sector, and a targeted visa program. » Develop capacity of technology transfer offices to convert MoESCS, Long term early stage research into market-ready technology and MoHTI complete deals. » Support development of local and diaspora-led business angel MoHTI, MoE Long term networks in Armenia. » Shift focus of publicly funded seed-funding programs (including MoHTI Short term incubators and accelerators) toward “successful exits” rather than “quantity of firms” in the program. » Introduce software patenting legislation to protect software MoHTI, MoJ Long term that demonstrates “further technical effect” (in line with EU guidelines). » Introduce managerial training programs to build middle- MoE, MoHTI    management capacity. Access to growth finance » Launch consultations and develop a fintech strategy to MoF, CBA Short term introduce innovative products that could help bridge the MSME financing gap. » Develop implementation strategy (including adequate MoF, CBA Short term regulatory safeguards and oversight) to use agents (e.g., the Armenian Post) for delivery of core financial services. » Expand targeted financial capability advisory services to MoF, CBA Long term MSMEs on financial products, financial reporting, financial management, business planning, and marketing to enhance their bankability and discourage informality. Table continues next page 34  |  Actions and Recommendations TABLE 4.1 (Continued) RECOMMENDATION AUTHORITY TIMEFRAME Corporate law and foreign direct investment » Introduce debt instruments (e.g., convertible loans) that can be SRLE, MoE, MoF Short term converted to equity in subsequent funding rounds. » Introduce corporate entities that support different classes of SRLE, MoE Short term voting shares (with differing entitlements to dividends and/or different levels of voting power) in order to better reflect early stage financing risks. » Develop clear rules pertaining to the fiduciary responsibilities of SRLE, MoE Long term corporate management teams toward minority shareholders (so that the latter are adequately protected—particularly where the interests of minority shareholders directly clash with those of majority shareholders). » Introduce legislation that allows investors to choose between MoE, MoF Short term local and international courts in the event of arbitration. » Strengthen investment “after care” by creating clear MoE, MoF Short term mechanisms to identify, track, and manage grievances between investors and public agencies before they escalate into disputes or withdrawals. » Repeal the 1994 Foreign Investment Law and enact new MoE, MoF Long term umbrella framework for investment in line with international best practice, including: ›Defining scope of application of the “fair and equitable treatment” principle, ›Defining characteristics of compensation payment, ›Defining limits of prohibition of direct and indirect expropriation, and ›Proscribing the rights of the investor to dispose of their profits and export their funds. » Build “investor roadmap” to identify permits, processes, and MoE, MoF Short term administrative changes that are constraining investments. » Design investment promotion strategy to attract global firms MoHTI, MoE, MoF Short term in niche sectors that can leverage Armenia’s particular factor conditions (e.g., biotech and cleantech). Table continues next page 35  |  Actions and Recommendations TABLE 4.1 (Continued) RECOMMENDATION AUTHORITY TIMEFRAME Digital trade » Expand the Law on Electronic Document and Electronic MoJ Long term Signature to recognize all electronic signatures and grant them, in addition, “presumption of validity and authenticity” in legal proceedings. » Include specific regulations for cybersecurity in the Law on MoJ Long Term Personal Data Protection, adding requirements such as the appointment of a data manager and risk assessment procedures. » Develop a framework that explicitly addresses online MoJ Long term intermediaries with safe harbor provisions to shield them from liability for infringing content posted by third parties, including intellectual property violation. » Work with the U.S. Department of Commerce and international MoE Long term counterparts to improve Armenia’s grouping in order to facilitate trade and collaboration in “dual-use” technologies as per the international Wassenaar Agreement. Data policy     » Implement policies relating to the classification, formatting, and MoHTI, MoE Short term storage of administrative and statistical data. » Implement policies that will enhance digital data management MoHTI, MoE Short term capacity and promote digitization among private firms. » Implement localization and data sovereignty laws that do not MoHTI, MoE Long Term stifle interest but still stimulate local economic activity in Armenia. » Provide technical assistance to help the private sector comply MoHTI, MoE Long term with the EU’s General Data Protection Regulations and meet international standards (e.g., the International Organization for Standardization–International Electrotechnical Commission 27000 family of standards). Education     » Integrate the planned High-Tech Industries Strategy with the MoESCS, MoHTI Short term planned Education, Science, Culture, and Sports Strategy. » Update bachelor’s, master’s, and PhD curricula in the fields of MoESCS, MoHTI Short term computer science and engineering to correspond with industry standards. Table continues next page 36  |  Actions and Recommendations TABLE 4.1 (Continued) RECOMMENDATION AUTHORITY TIMEFRAME » Introduce annual curriculum review process that uses public- MoESCS, MoHTI Short term private dialogue to solicit industry opinion and incorporates clear mechanism for revisions and updates. » Build an artificial intelligence PhD program and identify MoESCS, MoHTI Short term opportunities for joint research labs in collaboration with leading international research institutions (in both the public and private sectors) in this field. » Develop a quantum computing strategy and road map. This MoESCS, MoHTI Long term should include formal tertiary training in quantum computing at all three levels—bachelor’s, master’s, and doctoral. » Strengthen future incarnations of the Programme for Anti- MoF, MoE, MoESCS Short term Corruption Measures in Education as follows: › Provide adequate funding for coordination, monitoring, and implementation of all listed anti-corruption measures, › Reduce prevalence of short-term, noncompetitive appointments to increase employment security and predictability among university staff, › Introduce compliance and integrity into accreditation and reaccreditation criteria for higher education providers, and › Develop comprehensive corruption detection and reporting mechanisms in the higher education sector, in such a way as to protect whistleblowers. Note: CBA = Central Bank of Armenia; EU = European Union; MoE = Ministry of Economy; MoESCS = Ministry of Education, Science, Culture, and Sports; MoF = Ministry of Finance; MoHTI = Ministry of High- Tech Sectors; MoJ = Ministry of Justice; MSME = micro, small, and medium enterprises; SRLE = State Register of Legal Entities. 37  |  Actions and Recommendations APPENDIXES 38  | Appendixes Appendix A APPROACH Opportunities for better international integration and where and how value is created within the GVC so as to insertion into global value chains (GVCs) in various know which value-chain segments and activities to industries depend on a wide array of factors, including compete for. global demand trends, sourcing and localization, the In response, the World Bank Group has developed a strategies of multinational corporations, specific qualitative analytical approach to value-chain competi- competitive dynamics within each phase and strategic tiveness analysis that can better inform project design segment of the chain, and the type and rate of innova- and help Bank Group clients identify strategic options tion within the industry. Some of the factors at least to retain a greater share of value added from the global partially under a country’s control are the alignment of industries that they participate in. resource endowments and the provision of comple- This qualitative analysis goes beyond quantifying mentary public goods, with the required cost structures “how much” value is created by each country partici- and minimum efficient scale to compete in a specific pating in a GVC, and instead focuses on how value is phase nationally, regionally, or globally. created and by whom. It is based on concepts devel- Rodrik (2013) has observed recurring patterns of oped by Professor Michael E. Porter and implemented structural change, particularly in East Asia and other by the European Foundation for Cluster Excellence (in high-income countries in recent history. Economies association with the European Union and the Instituto have moved from agriculture and informal economic de Estudios Superiores de la Empresa Business School) activities where value addition is low to light manufac- in Europe and Latin America over the past 15 years. In turing (e.g., apparel, footwear, simple electronics), then incorporating this approach in its analytics, the World to heavy industries (e.g., automotive, machine tools), Bank Group is able to help clients leverage indus- and finally to services (e.g., finance, insurance) where try-specific public policy and investment options to productivity and value addition are found to be catalyze private sector development and foster the significantly higher. income growth of the 40 percent of the population with Policy makers have accordingly followed that linear the lowest incomes. logic to foster economic development. However, Rodrik (2014) asserts that globalization and the division of labor, global demand patterns, and the growing emphasis on technology and skills have radically altered that linear development path. Furthermore, the scattered nature of value-chain activities means that the capacity to capture value from a specific industry is determined by the policies of foreign competitors as well as domestic strategy. Not all value-chain activities can act as step- ping-stones to the next. Indeed, each value-chain activity has different potential to catalyze economic development and productive transformation. Therefore, as we move into a new, more uncertain phase of industrialization, it is important to understand 39  |  Appendix A Approach Appendix B METHODOLOGY STEP 1 STEP 3 GLOBAL INDUSTRY DEEP DIVE IDENTIFICATION OF PROTOTYPICAL VALUE CHAIN Background and contextual industry deep dive, includ- ing industry dynamics and trends related to global For each strategic segment identified in step 2, and supply (historical reason for success in certain coun- using similar resources: tries or regions, ways in which markets are responding to changes in demand), evolution in demand (geogra- a. An Advanced Buyer Purchase Criteria Assessment phy and product or service), industry cost structure focused on the minimum requirements (quality, and margins, recent evolution in value-chain structure volume, price, safety, transparency, traceability, (e.g., localization, specialization, and integration financial viability) that commercial buyers (both among firms). This work relies mostly on secondary business-to-business and business-to-consumers) research and secondary data. would accept. b. An assessment of geographic strategic options STEP 2 identified the advantages and disadvantages of STRATEGIC SEGMENTATION targeting local or global market using a single product or brand offering or strategy. Industry “segmentation” that groups each industry c. Key Success Factors—the characteristics that a firm not only by product or service but also by intended must have to compete well—were identified for market or user group using the global industry deep each strategic segment. dive in step 1, and Professor Michael E. Porter’s “five forces” analytical tool, which was also used to d. Resources a, b, and c were used to “map” the ideal assess industry attractiveness by determining the value chain for each strategic segment and identify: profitability of the industry and identifying the i. The optimal distribution of activities between actors within the industry with the most bargaining the local, regional and global levels of the power (thereby determining which actors capture value chain in terms of minimum efficient the bulk of the available profits). In addition, the scale of operation and production runs; analysis included some assessment of the evolution (historical and future) of this attractiveness. In ii. The type (knowledge-intensive, capital-inten- addition to traditional secondary resources (e.g., sive, labor-intensive, natural resources, and journals and papers), strategic segmentation relied energy-intensive) and level of intensity (low, on interviews with industry experts and nontradi- medium, high) of each value-chain activity; tional sources of information such as media (e.g., and online newspapers, magazines), annual reports, iii. Typology of value-chain linkages (just-in- filings, and blog posts. time, made-to-order, on-stock) by time (24 to 48 hours, weeks, months) and information exchange (high or low information exchange) that are required to compete effectively in a specific strategic segment. 40  |  Appendix B Methodology STEP 4 STEP 5 ARMENIA’S CURRENT POSITIONING BENCHMARKING AND FEASIBLE STRATEGIC OPTIONS A World Bank specialist traveled to Armenia and the United States to identify segments in which Armenia’s Strategic repositioning options for Armenia, including industries are currently participating. Consultations higher-value-added segment(s) that could be targeted; included almost 50 private firms and institutions in viable policy options to maximize value addition within Yerevan, San Diego, Silicon Valley, and Washington the segment(s); and necessary investments and D.C. Armenia’s performance in identified segments institutional changes that would be needed to support was assessed (compared to regional, as well as global, the development of Armenia’s productive capacity competitors) and the “ideal value chain” for those within the segment(s) were identified. Washington, segments was compared with the Armenian context. D.C.: World Bank Group. Specific policy, investments and/or institutional interventions that might be required for Armenia to better compete in those segments and capture more value were identified. 41  |  Appendix BMethodology NOTES 1. More information on the methodology and the 13. IP stands for intellectual property and, in this rationale for its selection is included in the appendix. case, is a throwback term to the earliest years of the semiconductor industry. 2. Virtually all the firms that were interviewed manufacture educational electronics toolkits spun out 14. In the semiconductor industry, “process of National Instruments after it had received specific nodes” or simply “nodes” are predefined technology requests for education solutions that were outside of development milestones. its core business. 15. Only five countries; Cuba, the Islamic 3. Homogenous genomes are easier to use as a Republic of Iran, the Democratic People’s Republic basis for research because “noise” arising from diverse of Korea, Sudan, and the Syrian Arab Republic are in ethnicities is greatly reduced, making results easier to Group E. There are 49 Group D countries; examples isolate and interactions easier to identify. include Afghanistan, Belarus, Central African Republic, the Russian Federation, and Zimbabwe. 4. Economists play an important role in cost-benefit analysis modelling. 16. June 24, 2019. 5. Natural or “pure” sciences use quantifiable, 17. Or impractical for a classical computer to empirical research methods to investigate natural solve as it will take thousands of years to finish. phenomena. 18. Modern encryption techniques rely heavily 6. Assuming (a) the distribution of bachelor’s on asymmetric encryption algorithms that would take degree programs in 2022 corresponds to the distri- centuries (or longer) to break using classical comput- bution of bachelor’s degree enrollments by program ing. However, in 1994, Peter Shor, an American math- in 2018, (b) the ratio of tertiary to bachelor’s degree ematician, wrote a quantum computer algorithm that graduates remains consistent, and (c) future tertiary could solve encryption algorithms at rates that are graduate figures are a function of current primary and exponentially faster. If the necessary quantum hard- secondary enrollment figures. ware were available, it would render current (asymmet- ric) encryption algorithms ineffective. 7. Yerevan State University’s master’s degree program in data science for business is implemented 19. Scimago Institutions Rankings database. jointly with the Innovative Solutions and Technologies 20. In April 2019, the United Nations Center and the Enterprise Incubator Foundation. Development Programme signed an agreement with Collaboration with the California State University of Granatus Ventures to help establish and manage the San Jose allows professors and researchers to integrate Granatus Tech4SDG Fund, aiming to raise US$40 with their U.S. counterparts and thereby better the million. research institute. 21. According to the European Patent Office’s 8. Integrated circuits are typically embedded on guidelines, “further technical effect” is going beyond a thin wafer (or “chip”) of semiconductor material— the “normal” physical interactions between the most often silicon. This is why they are interchange- program (software) and the computer (hardware) on ably referred to as silicon chips, semiconductor chips, which it runs. “semiconductors,” or “chips.” 22. “Digital signatures” fall under specific stan- 9. NASDAQ market capitalization on June 21, 2019. dards and stringent verification process. They explic- 10. Ibid. itly involve the use of a code or algorithm to sign and validate the authenticity of a document. 11. Mentor Graphics was acquired by Siemens in March 2017 for US$4.5 billion. 12. NASDAQ market capitalization on July 7, 2019. 42  | Notes REFERENCES ACN (Anti-Corruption Network for Eastern Europe Funding Report: Is Digital Health in a Bubble?” and Central Asia). 2018. Anti-Corruption Reforms in Rock Health, San Francisco. https://rockhealth.com/ Armenia: 4th Round of Monitoring of the Istanbul reports/2018-year-end-funding-report-is-digital-heal- Anti-Corruption Action Plan. Paris: Organisation for th-in-a-bubble/. Economic Co-operation and Development. https:// Gerbert, Philipp, and Frank Rueß. 2018. “Quantum www.oecd.org/corruption/acn/OECD-ACN-Armenia- Computing—and How to Play.” Boston Consul- -4th-Round-Monitoring-Report-July-2018-ENG.pdf. ting Group, Boston. https://www.bcg.com/en-us/ Altbach, Philip G., Liz Reisberg, Maria Yudkevich, publications/2018/next-decade-quantum-compu- Gregory Androushchak, and Iván F. Pacheco. 2012. ting-how-play.aspx. Paying the Professoriate. Routledge. https://www. Grand View Research. 2019. “Remote Patient Monito- bc.edu/content/dam/files/research_sites/cihe/pdf/ ring System Market Size, Share, and Trends IHEpdfs/ihe68.pdf. Analysis Report.” Grand View Research, San Arakelyan, Arsen, and Harry Glorikian. 2019. “The Francisco. https://www.grandviewresearch.com/ Armenian Genome Project Implmentation industry-analysis/ Strategy & Roadmap.” remote-patient-monitoring-devices-market. Basu, Diptasri. 2017. “The Electronic Health Records Green, Harriet. 2018. “IBM Sees Quantum Computing System In the UK.” Centre for Public Impact, Going Mainstream within Five Years.” Interview London. https://www.centreforpublicimpact.org/ by Hadley Gamble. Capital Connection, CNBC, case-study/electronic-health-records-system-uk/. March 30. Video, 2:37. https://www.cnbc. Batelle Memorial Institute. 2013. “The Impact of com/2018/03/30/ibm-sees-quantum-computing- Genomics on the U.S. Economy.” Battelle Memo- -going-mainstream-within-five-years.html. rial Institute , Columbus, OH. https://web.ornl. InfoDev. 2014. Building Competitive Green Industries: The gov/sci/techresources/Human_Genome/publica- Climate and Clean Technology Opportunity for t/2013BattelleReportImpact-of-Genomics-on-the- Developing Countries. Washington, DC: World Bank. -US-Economy.pdf. http://documents.worldbank.org/curated/ Bray, Freddie, Jacques Ferlay, Isabelle Soerjomataram, en/837201468165876160/pdf/91160-REPLACE- Rebecca L. Siegel, Lindsey A. Torre, and Ahmedin MENT-Building-Competitive-Green-Industries. Jemal. 2018. “Global Cancer Statistics 2018: pdf. GLOBOCAN Estimates of Incidence and Mortality IP/IT Committee for Armenia. 2019. “Boost Armenia’s Worldwide for 36 Cancers in 185 Countries.” CA: A Economic Development by Enhancing Innovation Cancer Journal for Clinicians 68 (6): 394–424. and by improving IP protection and enforcement American Cancer Society. https://doi.org/10.3322/ framework.” caac.21492. Lapedus, Mark. 2018. “Foundry Challenges in 2018.” Center for Applied Policy and Integrity. 2015. “Streng- Semiconductor Engineering, December 27. https:// thening Integrity and Fighting Corruption in semiengineering.com/foundry-challenges-in-2018/. Education: Armenia.” Report prepared on behalf Markets and Markets. 2019. “Geotechnical Instrumen- of the Open Society Foundations – Armenia, tation and Monitoring Market.” Markets and Center for Applied Policy and Integrity, Sofia. Markets, Pune. https://www.marketsandmarkets. http://www.osf.am/wp-content/uploads/2016/01/ com/Market-Reports/geotechnical-instrumenta- Integrity-report_final_en_12.11.2015.pdf. tion-monitoring-market-221101423.html. Coherent Market Insights. 2017. “Global Remote McCandless, David. 2017. “Codebases: Millons of Lines Patient Monitoring Devices Market.” Coherent of Codes.” Information is Beautiful. https:// Market Insights, Seattle. https://www.coherent- informationisbeautiful.net/visualizations/ marketinsights.com/market-insight/ million-lines-of-code/. remote-patient-monitoring-market-215. Medina, Leandro, and Friedrich Schneider. 2018. Day, Sean, and Megan Zweig. 2019. “2018 Year End “Shadow Economies Around the World: What Did 43  | References We Learn Over the Last 20 Years?” International https://www.researchandmarkets.com/research/ Monetary Fund, Washington, D.C. https://www. sllx8c/global_air?w=5. imf.org/en/Publications/WP/Issues/2018/01/25/ Sumant, Onkar, and Pallavi Jaiswal. 2016. “Remote Shadow-Economies-Around-the-World-What-Did- Patient Monitoring Market by Condition.” Allied -We-Learn-Over-the-Last-20-Years-45583. Market Research, Portland, OR. https://www. ON Semiconductor. 2019. “Implantable Medical alliedmarketresearch.com/ Devices.” ON Semiconductor, Phoenix. https:// remote-patient-monitoring-market. www.onsemi.com/PowerSolutions/content. Ugalmugale, Sumant, and Rupali Swain. 2019. “Digital do?id=16605&parentApp=16602. Health Market.” Global Market Insights, Selby- Pulidindi, Kiran, and Soumalya Chakraborty. 2019. ville, DE. https://www.gminsights.com/industry-a- “Seismic Reinforcement Materials Market.” Global nalysis/digital-health-market. Markets Insights, Selbyville, DE. https://www. WHO (World Health Organization). 2014. “Noncom- gminsights.com/industry-analysis/ municable Diseases Country Profiles 2014.” WHO, seismic-reinforcement-materials-market. Geneva. https://www.who.int/nmh/publications/ PWC (PricewaterhouseCoopers). 2017. “Semiconductor ncd-profiles-2014/en/. Industry: Strategies for Growth in the Internet of World Bank. 2013. “Armenia - BEEPS at a Glance 2013.” Things Era.” PWC, London. https://www.strate- World Bank, Washington, D.C. http://documents. gyand.pwc.com/gx/en/insights/semiconductor-in- worldbank.org/curated/en/699591468218377943/ dustry.html. Armenia-BEEPS-at-a-glance-2013. Rodrik, Dani. 2013. “Structural Change, Fundamentals, –––. 2017. Armenia - Systematic Country Diagnostic: and Growth: An Overview.” Institute for Advanced Sustainable, Inclusive Agriculture Sector Growth in Study, Princeton, NJ. http://j.mp/1HgaMHD. Armenia: Lessons from Recent Experience of Growth –––. 2014. “New Growth Strategies.” Keynote presenta- and Contraction. Washington, D.C.: World Bank tion at the World Bank Group’s headquarters, Group. “New Growth Strategies: Delivering on their –––. 2019. “Access to Finance - Technical Note.” World Promise?,” Washington, D.C., October 14–15. Bank, Washington, D.C. https://www.theciip.org/sites/ciip/files/documents/ Keynote%20-%20Prof%20Dani%20Rodrik.pdf. San Diego Regional EDC (Economic Development Corp). 2016. “Cracking the Code: The Economic Impact of San Diego’s Genomics Industry.” BW Research Partnership, Carlsbad, CA. https://www. sandiegobusiness.org/sites/default/files/ Cracking%20the%20Code%20-%20The%20 Economic%20Impact%20of%20San%20 Diego’s%20Genomics%20Industry_0.pdf. Schmaltz, Remi. 2017. “What is Precision Agriculture.” AgFunderNews, April 24. https://agfundernews.com/ what-is-precision-agriculture.html. SIPRI (Stockholm International Peace Research Institute). 2019. “SIPRI Military Expenditure Database.” SIPRI, Solna. https://www.sipri.org/ databases/milex. Sperling, Ed. 2017. “Is 7nm The Last Major Node?” Semiconductor Engineering, July 20. https://semien- gineering.com/7nm-last-major-node/. Stratistics Market Research Consulting Pvt Ltd. 2018. “Air Pollution Control Systems - Global Market Outlook (2017–2026).” Stratistics Market Research Consulting Pvt Ltd., Gaithersburg, MD. 44  | References Contact Ifeyinwa Uchenna Onugha ionugha@worldbank.org