By Shahid Yusuf
Hope springs eternal. The end of the Covid 19 pandemic is not yet in sight. Some fear that the second globalization is unravelling. 2021 was the sixth warmest year on record with the planet on track to exceed the 2°C limit by 2050 if not earlier. But techno-optimists are convinced that a digital transformation that will galvanize economies rich and poor, is within reach. There is no dearth of true believers. Key in the term digital transformation and Google registers nearly 4 billion hits, proof that it is attracting widespread attention.
Tom Siebel claims that the integration of cloud computing, Big Data, Artificial Intelligence (AI), and the Internet of Things (IoT) will be a game changer, infusing virtually all areas of the economy. That machine learning could prove to be a General Purpose Technology (GPT) is a claim voiced by Avi Goldfarb et al (2022) and M. Trajtenberg (2019). They maintain that it can push the technology frontier outward across a broad front, spurring a wealth of innovation and stimulating more creativity than destruction. The Nobel Laureate Daniel Kahneman (2019) is equally certain that AI/automation will improve mankind’s lot. “The robot will be much better at statistical reasoning and less enamored with stories and narratives than people are. The robot would have a much higher emotional intelligence…[And] the robot would be wiser…The essence of wisdom; is broad framing. A robot will be endowed with broad framing…. When it has learned enough, it will be wiser than we people because we…are narrow thinkers, noisy thinkers, and it is very easy to improve upon us. I do not think that there is very much that we can do that computer will not eventually be programmed to do.”
With the industrial revolution seemingly having run its course, a lot appears to ride on the success of digital technologies. Almost every country and many businesses are pinning their hopes on digitization to stage a speedy recovery from the Covid doldrums and to sustain growth thereafter. This note briefly address three questions: (i) Whether digitization can deliver the economic performance and income distribution that breakthrough technologies such as electricity, once did; (ii) What are the risks and constraints associated with rapid technological change; and (iii) How could middle-income countries extract the maximum economic mileage from digitization.
Question 1. There can be no doubt that the industrial revolution caused by technologies such as electricity and the internal combustion engine was truly transformative. It added a vast amount of value; it created an abundance of jobs; it generated a wealth of new products that dramatically improved living standards; it contributed to trade globalization; and countries that successfully industrialized, grew steadily, and achieved high levels of per capita income. Essentially, they established a benchmark for technologies to follow.
For close to two centuries, productivity augmenting industrialization was the well-trodden path to development. More recently, East Asian economies such as Korea, China, Taiwan, and Singapore demonstrated that the recipe worked. It was possible to prosper by building deep manufacturing capabilities and by exporting increasingly complex products. But from about 15 years ago, this reliable driver of growth began to lose steam. The share of manufacturing started declining first in the advanced countries and most steeply in lower middle-income countries like Pakistan (Figure 1). Academic publications began making references to premature deindustrialization. An inference drawn by some observers is that the future of development is more closely tied to services. And GDP growth would need to be sourced from improvements in productivity, with services playing the leading role, possibly displacing manufacturing.
Figure 1: Manufacturing value added in lower middle-income countries (% of GDP)
Source: World Development Indicators (2022) https://data.worldbank.org/indicator/NV.IND.MANF.ZS?locations=XN
It is because digitization is accelerating the shift away from industry and agriculture towards services and because the cross-border flows of data and information are rising fast, that the outcome of the transformation has taken on such significance.
The process of digitization broadly encompasses the use of computerization, robotics, automation, AI/machine learning, cloud computing, Big Data, algorithm development, and blockchain. Unlike earlier technologies, digitization has diffused at remarkable speed. Internet use has exploded with half the world’s population relying on it for communication, news, and information of every kind. Fourteen billion devices are now connected with the numbers increasing by the day. Since it was introduced in 2009, the smartphone has become nearly indispensable. At this rate, we will soon be surrounded by smart objects. Factories and farms in advanced economies are being automated and business processes upended. And it does not seem that digitization has peaked. With 5G being rolled out, autonomous vehicles on the near horizon, deep learning progressing in leaps and bounds, quantum computing coming within reach, and a steady trickle of innovations in neighboring fields reported weekly, digitization has plenty of momentum.
The tech visionaries believe that the digital era is in its infancy. They are convinced that it is only a matter of time before the new technology comes into full flower. After all, it took a few decades for electricity to make its mark on industrial productivity and innovation. Already, striking advances have been made in machine learning especially in the areas of perception and cognition. Firms that have harnessed the potential of AI platforms have used their algorithmic virtuosity and network effects to scale businesses, enlarge the scope of their activities and acquire competitive advantage across disparate industries. A McKinsey report estimated that Arab countries could add 4 percent to their GDP by fully exploiting digital technologies. India could create as many as 65 million new jobs by 2025. Digitization could boost Europe’s annual productivity growth by a percentage point, and by developing AI, it could potentially add $2.7 trillion to its economic output by 2030. Another report in 2019 claimed that digitization could add $13 trillion to global GDP by 2030.
Amazon is one example of a digital highflier. It began life as an on-line seller of books. Now it is not just an e-commerce giant offering a vast array of products and services, with Amazon Web Services it has morphed into a technology company. It is asserted that financial innovation (Fintech) could “change the landscape of financial services” and give rise to more inclusive financial systems. Ant Financial spun-off from Alibaba is an example of the power of AI algorithms when paired with a huge volume of data. Ant used the data generated by Alipay its mobile payments platform to successfully challenge incumbent firms in businesses ranging from wealth management to health insurance. With a tenth of the staff of the leading American banks, it serves ten times as many customers. And Ant is not alone. It has competitors in China such as Tencent (using WeChatPay) and similar firms have emerged in many countries. Alibaba backed Paytm in India had a 100 million mobile wallet users and nine million participating businesses in 2017. Coupang is Korea’s Amazon. Flipkart competes with Amazon in India.
Clearly, digital technologies have considerable potential. Few of us would want to wind the clock back to a time when we did not have the smartphones or an abundance of apps, Uber and AirBnB. The Covid pandemic brought home the advantages of online shopping, conducting business via the Internet, working from home, and staying connected whether by Zoom or WhatsApp. In short, digital technologies have conferred numerous advantages, some which GDP does not capture – what economists call consumer surplus.
The optimists are hopeful that measurable gains in productivity could materialize over the medium term. According to one study, Europe could increase its productivity by 20 percent in a decade through the widespread adoption of AI. All that it takes is for companies to make a few organizational changes. They need to invest in a suite of skills that complement digitization, acquire, and master the software, accumulate usable intellectual property, build a brand image, and harvest customer and market information – in other words, enhance the organization’s performance by focusing on intangibles. Intangible capital can unlock the potential of digital technology. Physical investment is still needed but in this new technological environment, it is assumed to play second fiddle to intangibles. According to a slew of academic papers, many leading firms are doing just that – the share of investment in intangibles has risen by almost 30 percent and American and British firms are already putting more of their money into intangible capital than tangible capital. It is claimed that firms that accumulate intangible assets grow faster because intangibles are more scalable, give rise to more spillovers, and generate greater synergies. And these firms, which comprise some of the top 20-30 percent in advanced economies, are more productive and profitable than firms that have been slow to invest in intangibles.
Question 2: As we leave the Covid pandemic years behind can we then look forward to decades of sustainable growth fueled by productivity, with ample employment opportunities for the workforce? The evidence suggests that thus far digitization is not moving key economic indicators and the medium-term outlook is cloudy.
Robert Solow a Nobel prize winning economist remarked that “you can see the computer age everywhere except in the productivity statistics”. That aphorism dates from the 1980s but it has current resonance. Digitization has been ongoing for more than three decades, but after a spurt from the mid 1990s, GDP growth has slowed. Most importantly, productivity growth, has declined in virtually all countries where new technologies have made deep sectoral inroads. This includes the US, Germany, Japan, Korea, and China. And there is no upturn in sight. The productivity of R&D has also dipped.
Past technological epochs, after a lag, generated numerous new occupations, job opportunities, and ladders offering income mobility. Attempts to forecast the likely impact of digitization point towards a substantial winnowing of jobs in all sectors. Between a quarter and one half of all jobs are at risk or at high risk in the United States, in Europe and in East Asia over the next two decades from automation and AI enabled technologies. Worldwide job losses could run into tens of millions.
Undoubtedly, continuing digitization will also create new jobs – in data analysis, machine learning, process automation, cyber security, software development, to service green infrastructures, and others still to be identified – many requiring specialized tertiary level technical skills and soft skills. But labor absorption into digital occupations and associated gains in productivity could be slowed by persistent labor supply bottlenecks.
In 2002, 40 percent of jobs in the US required medium level digital skills and only 5 percent called for high digital skills. By 2016, as the composition of industry and services changed, 48 percent of jobs required mid-level skills and 23 percent required advanced skills. But even in the US, too many workers fall short. A third of working age Americans are virtually devoid of digital skills and one in six cannot use email, web search or basic online tools. The survey of adult skills conducted by the OECD (PIAAC) also found that literacy, numeracy, and computer skills were equally scarce across the 33 countries surveyed (Japan and Singapore excepted). Labor market constraints are even more severe in middle and lower middle-income countries.
There are several risks that come with digitization. Let me mention four: long-term structural unemployment of those who cannot be absorbed into jobs that do materialize; rising income inequality; an increase in the market share of a few powerful businesses – the so-called superstar firms; and the vulnerability of digital infrastructure.
Rapid technological change could make it harder for new entrants with high school education or less to find jobs. Displaced workers especially ones who are older, less skilled, less mobile would also be at a disadvantage. The latter comprise most of the workforce in developed and developing economies. An increase in structural unemployment would be economically burdensome, and it would deepen political and social fissures. There are worrisome signs of such fissures in many countries.
Skill biased digital technologies can also widen income disparities. Already the so-called great divergence in incomes is a troubling development in several advanced economies and in South Asia as well. The share of income accruing to the top 10 percent of earners and the top 1 percent has been rising and could be exacerbated by the pandemic. There is concern that the middle class is being hollowed out and growth weakened. The inequality could become self-reinforcing because the well to do are able to provide their children with the expensive high-quality schooling needed to enter elite universities, learn the skills in high demand, and from there graduate into plum jobs.
A third risk is the increase in market power of a few large firms that utilize digital platforms to dominate key industries. This is a development that can stifle competition, new entry, and innovation – and by concentrating wealth and political power, can be difficult to reverse.
A fourth concern is the increasing vulnerability of a digitized and networked economy and society to cyber-attacks, to weaponization of key networks, to severe weather events and misinformation. A smart city could be brought to its knees if key infrastructures were damaged or disabled. Essential services would be disrupted, and automated factories would grind to a halt. Heavy rain that inundated a Chinese city in 2021 provided a glimpse of what can happen. It was responsible for a digital dark age with cellphone services interrupted and residents unable to access ATMs or use their smartphone apps to pay for purchases.
Question 3: What then are the messages for Pakistan and other lower middle-income countries?
Five are uppermost. First it may be wise to make haste slowly – investing in resilient digital infrastructure and skills but balancing these with other priorities. Attempts to digitize rapidly are likely to encounter problems of technology absorption, there are upfront costs, and few medium-run economic benefits. Second, a selective adoption of ‘human friendly’ digital offerings is in order, tailored to the evolving capabilities of the workforce. A rush to embrace labor displacing, skill-intensive technologies, and the Internet of things (IoT) will be capital intensive, squeeze employment opportunities, depress the share of wages, and leave growth unchanged.
Third, a message that is frequently repeated, raising the quality of human capital through education and training will be necessary to capitalize on useful, productivity enhancing technologies. To get good jobs, people will need higher order skills. But accumulating relevant skills will take years and retraining current labor cohorts with weak or non-existent foundational skills, will be difficult. Improving and complementing post-secondary school education with targeted vocational training may be the way forward. A strong start and sustained focused effort, which is adequately funded could begin to pay-off in a decade or two.
Fourth, a competitive market environment with ease of entry and churning at the top is more likely to boost productivity and innovativeness than one that is dominated by a few oligopolies whether they are in private hands or owned by the state. With digital technologies contributing to business concentration in Europe, the US and China, regulators are waking up to the risks and beginning to act.
Fifth, I tend to discount the scope for services led growth irrespective of how it is juiced up with digital technologies and intangibles. Tangible and intangible capital must coexist and complement each other. Absent tangible productive capital, the accumulation of intangible capital even if it were to occur, will generate meager returns. No country has demonstrated that a services centric economy focused on intangibles has achieved or sustained rapid growth of the kind Pakistan needs. The handful of credible success stories based on East Asian experience all underscore the role of massive investment in physical capital – industry, infrastructures, and urban development – complemented by gradual improvement in the quality and volume of human capital. The literature on the ‘East Asian miracle’ makes no mention of intangibles. Middle-income countries – must hedge their bets and focus equally on building more complex manufacturing capabilities, which can add value by leveraging services.
The pattern of demand during 2020-2021 showed that demand for manufactures is as strong as ever. A mix of servitized manufacturing and digitized tradable services may be the path to longer term growth – but this will take investment in physical, human and knowledge capital. An inclusive digital transformation will take work. It is not around the corner.
Shahid Yusuf is chief economist at the Growth Dialogue and Non-resident fellow at the Center for Global Development both in Washington DC. He was formerly on the staff of the World Bank.
 Chief Economist, Growth Dialogue; and Non-Resident Fellow Center for Global Development, Washington DC.
 T. Siebel (2019) Digital Transformation: Survive and thrive in an era of mass extinction. Rosetta Books.
 A. Goldfarb, B. Taska and F. Teodoridis (2022) ‘Can machine learning be a GPT?’ NBER. https://www.nber.org/papers/w29767; M. Trajtenberg (2019) AI as the next GPT. Their views are largely echoed by contributors to the NBER conference on AI. A. Agrawal et al eds. (2019) The economics of AI: An Agenda. U. of Chicago Press. https://www.nber.org/books-and-chapters/economics-artificial-intelligence-agenda/artificial-intelligence-next-gpt-political-economy-perspective
 D. Kahneman (2019). Comment in, The economics of AI. https://www.nber.org/system/files/chapters/c14016/c14016.pdf
 D. Rodrik (2015) ‘Premature deindustrialization’. J. of Economic Growth. https://drodrik.scholar.harvard.edu/files/dani-rodrik/files/premature_deindustrialization.pdf; based on recent research, Rodrik states (2021) “Manufacturing is not the growth escalator it once was…Since 1990, practically no country outside of East and Southeast Asia has managed to reach or sustain employment levels in manufacturing exceeding 20 percent of the labor force, with the vast majority of developing nations falling far short of this threshold…Middle-income countries are experiencing declines in manufacturing employment shares at much lower levels of industrialization and of per-capita GDP, while low-income countries are finding it virtually impossible to replicate the experience of previous generation of manufacturing success stories. Moreover, in the few low-income countries where industrialization seems not to have run out of steam, its quality is very poor… Employment growth in these relative success stories seems limited to unregistered/informal parts of manufacturing, with formal manufacturing still remaining in the grasp of premature de-industrialization.” ‘Prospects for global economic convergence.’ https://drodrik.scholar.harvard.edu/files/dani-rodrik/files/prospects_for_global_economic_convergence_under_new_technologies.pdf; H. Kruse et al (2021) ‘A manufacturing renaissance?’ https://www.wider.unu.edu/publication/manufacturing-renaissance-industrialization-trends-developing-world
 The World Bank (2021) has hailed this structural change and maintains that it holds much promise. At your service. https://www.worldbank.org/en/topic/competitiveness/publication/promise-of-services-led-development; The optimism is echoed by R. Newfarmer et al (2020) Industries without smokestacks. https://www.wider.unu.edu/publication/industries-without-smokestacks-2; UNCTAD (2021) Digital Economy Report 2021. https://unctad.org/system/files/official-document/der2021_en.pdf; Mckinsey (2016) Digital globalization. https://www.mckinsey.com/business-functions/mckinsey-digital/our-insights/digital-globalization-the-new-era-of-global-flows
 Some of the trends and advances are presented in the AI Index Report 2021. https://aiindex.stanford.edu/report/; https://www.technologyreview.com/2021/01/14/1016122/these-five-ai-developments-will-shape-2021-and-beyond/; D. West and J. Allen (2018) How AI is transforming the world. Brookings. https://www.brookings.edu/research/how-artificial-intelligence-is-transforming-the-world/
 G. Westerman et al (2014) Leading Digital. Harvard Business. D. Rogers (2016) The Digital Transformation Playbook. Columbia.
 McKinsey (2016) Digital Middle East https://www.mckinsey.com/~/media/mckinsey/featured%20insights/middle%20east%20and%20africa/digital%20middle%20east%20transforming%20the%20region%20into%20a%20leading%20digital%20economy/digital-middle-east-final-updated.pdf
 McKinsey (2019) Digital India. https://www.mckinsey.com/business-functions/mckinsey-digital/our-insights/digital-india-technology-to-transform-a-connected-nation These projections predate the Covid pandemic.
McKinsey (2019) Innovation in Europe. https://www.mckinsey.com/~/media/mckinsey/featured%20insights/innovation/reviving%20innovation%20in%20europe/mgi-innovation-in-europe-discussion-paper-oct2019-vf.pdf
 McKinsey (2019) Twenty-five years of digitization. https://www.mckinsey.com/~/media/mckinsey/business%20functions/mckinsey%20digital/our%20insights/twenty-five%20years%20of%20digitization%20ten%20insights%20into%20how%20to%20play%20it%20right/mgi-briefing-note-twenty-five-years-of-digitization-may-2019.ashx
 T. Beck and Y-C Park (2021) Fostering Fintech for financial transformation. VoxEu. https://voxeu.org/content/fostering-fintech-financial-transformation-case-south-korea; https://voxeu.org/article/fostering-fintech-financial-transformation
 A. Collis (2019) How should we measure the digital economy? HBR. https://hbr.org/2019/11/how-should-we-measure-the-digital-economy; It is claimed that conventional measures of GDP are missing “dark matter”. “GDP captures the value of all goods and services produced in the U.S. economy, as measured in the dollar value of all products sold. That means non-paid transactions like YouTube views, Facebook browsing, Netflix bingeing, and Google searches don’t register (except to the extent that they generate more advertising revenue).” There is truth to this contention however, the value of the missing dark matter may be relatively small. T. Hyde (2017) Is digital dark matter skewing our GDP statistics? https://www.aeaweb.org/research/is-digital-dark-matter-skewing-gdp-statistics
 J. Haskel and S. Westlake (2022) https://www.mckinsey.com/business-functions/marketing-and-sales/our-insights/forward-thinking-on-the-transformative-role-of-intangible-assets; (2017) Capitalism without Capital. Princeton University Press; C. Corrado, C. Hulten, D. Sichel (2006) https://www.federalreserve.gov/pubs/feds/2006/200624/200624pap.pdf; Financing intangible capital is an issue. L. Demmou and G. Franco (2021) Mind the financing gap. https://voxeu.org/article/enhancing-contribution-intangible-assets-productivity
 McKinsey (2021) https://www.mckinsey.com/business-functions/marketing-and-sales/our-insights/getting-tangible-about-intangibles-the-future-of-growth-and-productivity; https://www.caixabankresearch.com/en/economics-markets/activity-growth/intangibles-new-investment-knowledge-era; https://www.weforum.org/agenda/2021/07/defining-intangible-investments-grow-knowledge-economy/; https://voxeu.org/article/intangibles-and-productivity-growth-evidence-japan-and-korea; https://voxeu.org/article/productivity-and-secular-stagnation-intangible-economy; “By 2013, for every £1 of investment in tangible assets (buildings, machines, vehicles, etc.) the major developed countries spend £1.10 on intangible assets (such as software, R&D, design, branding, training, and business process engineering). We ask whether this move to intangibles helps explain secular stagnation.”
 D. Acemoglu and P. Restrepo (2020) estimate that “each robot displaces 3.3 jobs.” Robots and jobs. JPE. https://economics.mit.edu/files/19696; Oxford Economics (2019) went on to add, “More than half the workers who left production jobs in the past two decades (in the United States) were absorbed into just three occupational categories: transport, construction and maintenance, and office and administration work. Ominously, our analysis found that these three occupational areas are among the most vulnerable to automation over the next decade”. https://cdn2.hubspot.net/hubfs/2240363/Report%20-%20How%20Robots%20Change%20the%20World.pdf?utm_medium=email&_hsenc=p2ANqtz–S_yv5LZTWzdC5IER_NtSl3PcknlmRKCRLWkiY7DXoc24tLeHNQmxbfIluLCA4PrkWMen4_J_hWSH49WG3OQvHF61Jlg&_hsmi=74013545&utm_content=74013545&utm_source=hs_automation&hsCtaTracking=07b1855a-24f4-4b99-bcb8-b0d2a13b715e%7C53b7a48e-9591-4179-8eab-694443190b4f; However, thus far the use of robots is concentrated in a few manufacturing industries and in logistics. E. Benmelech and M. Zator (2022) use data from Germany to show that investment in robots is modest, generally in response to labor shortages and firms that do invest increase employment. Robots and firm investment. NBER. https://www.nber.org/papers/w29676
 OECD (2016) Skills Matter. https://read.oecd-ilibrary.org/education/skills-matter_9789264258051-en#page22
 A study by CGD found that the education quality has stagnated or declined in South Asia and Sub-Sharan Africa. A. Le Nestour et al (2022) The long-run decline of education quality in the developing world. https://www.cgdev.org/publication/long-run-decline-education-quality-developing-world
Billionaires getting richer; the better educated being able to continue working from home; rising asset and home prices. F. Ferreira (2021) https://www.imf.org/external/pubs/ft/fandd/2021/06/inequality-and-covid-19-ferreira.htm; ADB (2019) Demystifying income inequality in Asia. https://www.adb.org/publications/demystifying-rising-inequality-asia; B. Perry (2020) https://www.pnas.org/doi/10.1073/pnas.2020685118; N. Angelov (2021) https://voxeu.org/article/income-inequality-during-covid-19-pandemic; https://www.aljazeera.com/economy/2021/10/12/japan-faces-rising-inequality-after-eight-years-of-abenomics; https://www.theglobaleconomy.com/Germany/income_top_10_percent_earners/
 Z. Qureshi (2019) https://www.brookings.edu/blog/up-front/2019/05/21/the-rise-of-corporate-market-power/; Big firms are more profitable but less productive. Gutierrez and Philippon (2019) https://www.aeaweb.org/articles?id=10.1257/pandp.20191065
 Factory automation, on-line shopping, self-checkout kiosks, automated customer service, even Fintech, is not spurring productivity growth in countries with technological capabilities and an adequate bench of skills. Productivity growth in China, which has taken an early lead in Fintech has slumped over the past decade. L. Brandt et al (2020) China’s productivity slowdown and growth prospects. https://openknowledge.worldbank.org/handle/10986/33993
 In most advanced countries between 20 and 50 percent of students receive vocational training via formal institutions and apprenticeship programs. The college focused approach adopted by the US has proven to be exceptionally leaky. Of 100 high school students, 13 will not complete high school, 29 will complete but not go on to college. 27 will enroll but fail to get a degree. Of the 31 earning a college degree, 13 will be in jobs not requiring one and only 18 go from high school to college to career in six years. https://nscresearchcenter.org/completing-college/
 “American and world technology is shaped by the decisions of a handful of very large, very successful tech companies that have tiny workforces and a business model built on automation. Big Tech companies including Amazon, Alibaba, Alphabet, Facebook, and Netflix are responsible for more than $2 of every $3 spent globally on AI. Their vision, centered on the substitution of algorithms for humans, influences not only their own spending but also what other companies prioritize and the aspirations and focus of hundreds of thousands of young students and researchers specializing in computer and data sciences.” D. Acemoglu (2021) Remaking the post-Covid world. https://www.imf.org/external/pubs/ft/fandd/2021/03/COVID-inequality-and-automation-acemoglu.htm
 Panama and Dubai are the exceptions but their experience is of little relevance for Pakistan and other lower middle income countries.