Source: Authors’ elaboration based on UNDP Global Multidimensional Poverty Index.

Foreign investment, combined with local information technology (IT) and engineering
expertise, produced masses of new good jobs and triggered the growth of a thriving middle
class. In turn, this new mass domestic market primed the growth pump by attracting more
foreign investment and stimulating job and firm creation that rose to meet the rise in Indian
consumer demand.

These reforms helped the overall growth rate to rise from its sub-4% performance for most of
the post-independence period. In recent years, India has become one of the great growth
success stories of the 21st century, lifting hundreds of millions of people out of abject poverty.
From the 1980s, and especially from the 1990s, India’s growth shifted into higher gear. Now,
an annual growth rate of over 8% is expected by many (see Figure 7), and poverty rates have
dropped by half (from 55% to 28%) between 2005 and 2015 according to UNDPs
multidimensional poverty index.

Note that the connection between the reforms in the 1980s and 1990s, and the resulting
growth take-off has been questioned by Rodrik and Subramanian (2005). They argue that the
growth was not the outcome of economic or policy changes, but rather due to social-
psychological factors that they call ‘an attitudinal shift’ toward a pro-business stance. As they
put it: ‘the trigger for India’s economic growth was an attitudinal shift on the part of the
national government … that unleashed the animal spirits of the Indian private sector’.
Panagariya (2004) disputes their conclusions, arguing for a more traditional policy-linked
explanation, where trade liberalization and relaxation of industrial controls had a major role.

The service sector and export boom

The post-reform growth pattern, however, was not what many were expecting. The received
wisdom was that adopting the ‘Washington Consensus’ reforms would unleash rapid
industrialization and manufacturing-led-export growth. In India, however, it was the services
sector that led the way (Murthy 2005; Nayyar 2012; Basu 2015). Conventional thinking
about development, especially trade and development, has largely ignored services in general
and service exports in particular, so this outcome was unexpected.

1
The canonical treatment about what we should expect from services in the course of
economic development is highlighted in the study by Eichengreen and Gupta (2009). They
document two distinct phases where the service sector is important in development. First,
when an economy moves from low-income to middle-income, various informal services
sectors grow rapidly. Second, when it moves beyond middle-income, more sophisticated
services sectors become important, such as information technology and finance.

India’s post-1991 growth performance was, in essence, the early arrival of the second phase
in the Eichengreen-Gupta story. Basu (2015) describes it as: ‘this second-stage services
sector growth that happened in India, rather early and with a vigor rarely seen anywhere
else’. Even more intriguing was the fact that the leading service sectors were those of the
‘knowledge economy’—a set of activities that were ‘traditionally viewed as the preserve of
advanced economies’. But perhaps the outcome should not have been so unexpected given
the constraints on Indian manufacturing production and trade.

The Indian education sector has long favored high-quality universities, especially in science,
technology, and engineering disciplines. As a result, it presented an abundance of well-
trained and talented technological workers. However, while this might have fostered
manufacturing as it did in Germany and Switzerland, the policy environment did not allow it.
And while reforms removed some of the barriers, key limitations continued to restrict Indian
manufacturing. A poor transportation infrastructure as well as a great distance to the
manufacturing giants (US, Germany, Japan, and China) tended to shelter the Indian market
from foreign competition. And while this might have fostered production, it also made India
an unlikely participant in the global value chains that had become essential to
competitiveness since the late 1980s.

Services, by contrast, were largely unaffected by these constraints. The service sector,
especially the IT services sector, was untouched by transportation issues, faced no explicit
import or export barriers and in addition remained largely untaxed. In short, what Basu
(2015) calls ‘India’s overproduction of engineers throughout the 1960s, 1970s, and 1980s’
matched Silicon Valley’s booming demand for tech workers (and the US’s underproduction
of engineers).

For a variety of reasons (see Weiner 1991), India spent, and still spends, relatively far more
resources on higher education than on primary education. In 2000, the country spent 86% of
per capita GDP on each student in tertiary education, compared to 14% of per capita GDP per
student in primary education. By contrast, China was spending 10.7% and 12.1% of per
capita GDP per student in tertiary and primary education respectively. And the Indian
students coming through this system were world-class, receiving over 50% of US non-
immigrant work visas requiring specialized skills.

While it is impossible to identify the causes with certainty, the outcome was plain for all to
see. India had become a magnet for IT and knowledge-based jobs that were offshored from
advanced economies, and Western firms were shifting their research and development
activities to the country in order to reduce costs and avoid talent shortages. India was
becoming host to outsourced call centers and many so-called business process outsourcing
(BPO) activities, including many back-office jobs like medical billing, business
administration, and labor-intensive insurance-related services. And this rapid development of
the sector was accompanied by the rise of several world-beating multinationals like Infosys,
HCL, and Wipro.

2
5.2 The Philippines

‘I couldn’t do my job without her and I tell her that’, relates Alison De Kleuver when she was
finance and operations director for the global company EY. Referring to Sandy, her
Manilabased remote assistant, De Kleuver continues: ‘She helps to optimize my time, both in
what she does for me but also in organizing my schedule, and that is invaluable to me’
(Tadros 2018). And it is cheaper. ‘The key number we’re finding with this model is that an
offshore [executive assistant] costs approximately 40% of an onshore executive assistant’,
according to De Kleuver. That is a big saving as the firm used to spend between US$50,000
and US$95,000 for an executive assistant in Sydney or Melbourne. This is not an unusual
story.

The Philippines has also reaped the benefits of the technological-led freeing of service-sector
trade. ‘Rarely has a new industry traced the trajectory from concept to prime economic driver
as quickly as business process outsourcing (BPO) has in the Philippines’, noted a report by
the consultancy Oxford Business Group (2016). The sector, in particular the call center
industry has seen stellar growth, soaring from 2,400 employees across just four operations in
2,400 in 2000 to 375,5000 employees across 425 call center operations in 2012 (Chang and
Huynh 2016).

In 2000, the IT-based business process outsourcing (IT-BPO) sector was contributing almost
nothing to the nation’s GDP. By 2016 it had accounted for about 7% of GDP and was
employing over 1.1 million; 1,000% more than in 2004 (IBPAP 2016). Indeed, IT-BPO is the
country’s largest private employer of white-collar workers. The industry projects the creation
of over 600,000 new jobs in the sector by 2022. And these are good jobs, with IT-BPO
employees typically earning twice the national average in the Philippines.

A bit on the history

So, why the Philippines? The answer lies partly in factor ‘endowments’ and partly in
government policy. The international BPO industry was attracted by the country’s youthful,
literate population, armed with good English-language communication skills. Some analysts
also cite Filipinos’ strong customer-service orientation as well as the country’s adaptability to
consumers’ Western culture as an advantage (IBPAP 2016). It also helped that the
government set up special economic zones in 1995 that provided tax incentives to call center
operators.

While this push was led by call centers, the availability of talented, low-cost service workers
also attracted multinationals interested in setting up ‘shared services offices’. These are
offices where a variety of the company’s operations, such as accounting, payroll, human
resources, and various legal and IT services are clustered together and offshored to reduce the
wage bill. Companies engaging in these activities include HSBC, Standard Chartered, Capital
One, Citibank, and JP Morgan as well as Accenture, Oracle, Microsoft, and Dell.

Originally, the industry was initially clustered near Manila but it has more recently started to
spread further afield, thus stimulating regional economic development in the Philippines. The
industry is expanding into what they call Next Wave Cities such as Baguio, Bacolod, and
Cagayan de Oro.

International freelancing, while not as large as BPO activities, is also important. While there
is no official data on freelancing, a recent PayPal survey suggests that freelancing is booming

3
in the Philippines. The firm estimates that there are approximately 1.5 million freelancers in
the country and most of them were expecting to continue freelancing in the future. Many of
them are young, with 90% of respondents under 40 years old. 61% were women. In terms of
services provided, about a third engaged in data entry or internet research, and a large share,
about 60%, were working for US-based clients. Most lined up work via internet freelancing
platforms such as Upwork.com, Freelancer.com, or Shutterstock.com (PayPal Global
Freelancer Insights Report 2018 as cited in Esmael 2018).

Policy support

The country initially launched into labor-service exports via call centers, where Filipinos
answered phone-in questions. Since then, the Philippines has experienced a very clear
‘upgrading’ in their labor-service exports and the country’s activities are moving up the
skill/wage scale into what is known as knowledge process outsourcing. This includes back-
office services such as health-care processing and coding, legal transcription, IT outsourcing,
and more recently, animation and game development. The sector is increasing the provision
of high-end services such as data analytics, business and financial research, mortgage
servicing, software development, legal process and patent research as well as engineering. It
is estimated that there are already 200,000 Filipinos working in these higher-paid knowledge
process outsourcing jobs (Oxford Business Group 2016).

The IT-BPM Roadmap 2022 (IBPAP 2016), classifies these jobs under four main headings
with a number of sub-headings, namely:

• Contact Center and BPO subsector o

Engineering Services Outsourcing (ESO) o
Data Analytics o Performance Management o
Legal Process Outsourcing (LPO)
• Information Technology (IT) Services
subsector o Application Development
Management (ADM)
o System Integration o
Automation Enablement o
IoT-Enablement languages
• Health Information Management (HIM)
subsector o Preventive Health o Remote
Healthcare Management o Provider Services
• Animation and Game Development subsector o
3D animation o Augmented & Virtual Reality
(AR/VR) o Gamification

This move into higher-value-added IT-based services depended critically on two key laws:
the Data Privacy Act of 2012, which established penalties for unauthorized use or disclosure
of personal information, and the Cybercrime Prevention Act of 2012, which set up a legal
framework to identify, prevent and punish cybercrime.

Education policy has also been supportive. The ‘machinery’ of IT-BPO are not machines, but
humans, and education and training programs are critical for the continued expansion and
evolution of the industry. Noting this, the leading industry association, IBPAP, worked with

4
the government’s Commission on Higher Education to maintain the flow of skilled service-
sector workers. One notable output of the cooperation was the Service Management Program,
which offers specialized courses for students of business administration, management, or IT,
with the goal of placing them in entry-level IT-BPO positions. The program also funds a
‘train the teachers’ initiative, again focusing on teaching IT-BPO subjects at the industry’s
standards. The effort has spread across the country and now includes 17 state universities and
colleges.

The future is looking bright, according to the IT-BPM Roadmap 2022. By 2022, the sector is
projected to account for 7.6 million direct and indirect jobs—a half million of which would
be outside the Manila area. Of these, a million jobs are projected to be in higher-value areas.
To support this growth, the government created a Department of Information and
Communication Technology.

In terms of policies to support future growth, the industry has concentrated on three action
areas: (1) widening and deepening human capital by scaling up industry public-private
partnerships; (2) bolstering the attractiveness of the Philippines as an investment destination
through advocacy activities; and (3) building the Philippine IT-BPM brand globally through
marketing programs.

Automation poses challenges

Globally, automation is creating displacement in both the service and manufacturing sector.
While the biggest impact is likely to be in advanced economies, the trend is affecting the
Philippine service-export sector. A global network of shared-service and outsourcing
professionals, called SSON, has produced a report that looks into some of the challenges
(Shared Services and Outsourcing Network 2018). It notes: ‘While local Shared Services
providing global enterprise support will continue to play their role, there is no denying the
signs that robotic process automation (RPA) is emerging as a solution that will impact the
offshore services equation.’ RPA is a widespread form of service-task automation that is
spreading to offices around the world. It allows computers to take over some ‘knowledge
assembly-line’ tasks that used to require humans until machine learning became
commercially viable.

Some of the Philippines-based companies in the sector have started to embrace RPA
solutions as part of their services offering to advanced-economy firms. For example, they can
offer RPAbased automation with humans that can deal with the inevitable exceptions that the
RPA cannot manage. One consulting firm active in the area relates the example of Philippine
outsourcing firms leveraging chatbots, machine learning, and natural language processing in
the contact centers to enhance the capacities and productivity of their human workers in
handling customer interactions (Karthik and Kala 2019).

5.3 The Chinese case

The People’s Republic of China was founded on 1 October 1949 by the Communist Party of
China under the leadership of Mao Zedong, ending the long chaotic period that followed the
1911 overthrow of the Qing dynasty. Industrialization was shallow in 1949 and remained
largely a coastal phenomenon. In 1952, the secondary sector produced 8% of GDP and
employed 7% of the labor force compared with the primary sector, which produced 74% of
GDP and employed 84% of the workforce. The coastal provinces retained 72% of fixed
assets and accounted for 69% of the gross value of industrial output (Yang 1997). Naturally,

5
the Communist Party at its accession to power in 1949 regarded industrialization as its most
important economic task .

A large share of the country’s economic output was directed and controlled by the state,
which set production goals, controlled prices and allocated resources throughout most of the
economy. During the 1950s, all of China’s individual household farms were collectivized
into large communes. To support rapid industrialization during the 1960s and 1970s, the
central government undertook large-scale investments in physical and human capital. As a
result, by 1978 nearly 75% of industrial production was in the hands of centrally controlled,
state-owned enterprises that followed centrally-planned output targets. Private enterprises and
foreigninvested firms were generally barred.

A central goal of the Chinese government was to make China’s economy relatively self-
sufficient and foreign trade was generally limited to obtaining those goods that could not be
made or obtained in China. The State Planning Commission’s import plan covered more than
90% of all imports, with the export plan being similarly comprehensive, specifying the
physical quantities of more than 3,000 individual commodities. Prior to 1978, a handful of
foreign trade corporations owned and controlled by the Ministry of Foreign Trade were
responsible for carrying out the import and export plans. The volume of Chinese trade,
relative to world trade, declined sharply from 1.5% in 1953 to 0.6% in 1977 (Lardy 1994: 2).

In 1978, things changed. China decided to break with its Soviet-style economic policies by
gradually reforming the economy and sequentially opening up trade and investment with the
West, in line with free-market principles, and in the hope that this would significantly
increase economic growth and raise living standards. As Chinese leader Deng Xiaoping, the
architect of China’s economic reforms, famously put it: ‘Black cat, white cat, what does it
matter what color the cat is as long as it catches mice?’

(Growth accounting reveals that) Much of China’s rapid economic growth after 1979 can be
attributed to two main factors: large-scale capital investment, financed by large domestic
savings and foreign investment), and rapid productivity growth (Bosworth and Collins 2008).
This coincided with a massive reallocation of labor from agriculture to non-agriculture, made
possible by a green revolution in agriculture that sharply increased productivity in this sector.
The share of the labor force in agriculture fell from 75% in 1977 to 33% in 2012, while the
share of valueadded produced in the agricultural sector fell from 30% to 5%.

China also gradually reformed its trade regime and these reforms led China’s foreign trade to
soar from US$21 billion in 1978, when China was, at best, a marginal player in global trade,
to more than US$2.2 trillion today when China has become the world’s largest exporter
(National Bureau of Statistics 2005: 161; World Bank Data).

The Open Door Policy

An important piece of China’s trade reforms was the Open Door Policy that consisted of
attracting foreign direct investment and promoting foreign trade in targeted areas. This
opening up was initially limited to two southern provinces (Guangdong and Fujian), and then
gradually was extended to larger geographical regions: first along the coast and then to the
inland provinces. The open economic zones provided investors with various preferential tax
treatments and exemptions on duties and from labor regulations. The leading role of this
selective open door policy in regional growth has been emphasized by a great number of
studies (e.g. Mody and Wang 1997; Berthélemy and Démurger 2000; Chen and Feng 2000)

6
Foreign direct investment inflows did not occur immediately in large volumes in response to
the establishment of special economic zones in Guangdong (1979) and in Fujian (1980),
partly out of caution and partly because the liberal regulatory framework only began to be
introduced in 1982. Foreign direct investment started pouring in only from 1984 onward
(when it doubled from US$0.6 billion in 1983 to US$1.3 billion in 1984). The second large
acceleration of foreign direct investment inflow occurred in 1992, expanding from US$4.4
billion in 1991 to US$11 billion in 1992.

China’s international trade expanded steadily along with China’s share in global trade. The
export basket diversified from light manufacturing to heavy manufacturing and electronics.
Global value chains in the country expanded rapidly, starting in the early 1990s—a trend that
was accelerated by the lock-in provided by China’s World Trade Organization membership
in 2001.

6. Telemigration

Companies in G7 nations are turning to remote workers to perform an increasingly wide

range of tasks. For the most part, these remote workers are in the same nation as the
companies. Hence, it is less globalization and more wage differences and talent shortages that
are driving an increasing number of companies to turn to foreign-based online service
workers, or ‘telemigrants’ (Baldwin 2019). A recent study of Upwork contracts found that the
top three nations hiring telemigrants were the high-wage English-speaking nations of US,
Australia, and the UK. Conversely, the three biggest sources of telemigrants were the
Philippines, India, and Bangladesh (Horton et al. 2017). The US was the only high-wage
nation that was both a major buyer and seller of this sort of online remote labor.

When it comes certain service professions such as accountants, computer programmers,

engineers, nurses, and many others, complete replacement of a domestic worker with a
telemigrant would be impossible. However some substitution of low-cost foreign remote
workers for high-cost domestic workers would surely save money.

Who are today’s foreign freelancers? The online payments company, Payoneer.com, queried
23,000 freelancers worldwide. About 25% respondents were in Latin America and Asia, 20%
in Central and Eastern Europe, and about 15% in both the Mideast and Africa (Sukman
2015). The vast majority of freelancers surveyed are in their 20s and 30s (about 85%). A bit
more than half had university educations. The companies paying for their services were about
half in North America and Europe (split equally), about 15% in both Latin America and Asia,
and 7% in Australia and New Zealand.

How fast will telemigration grow? The answer will differ across the various types of service
sectors since some lend themselves much more easily to integrating remote workers or have
more widely accepted standards. Government regulation will also surely play a large role.

6.1 Factors driving telemigration

There are four factors suggesting that telemigration will grow faster than most think across
almost all sectors. Perhaps the most remarkable is how fast digitech is lowering the language
barrier.

7
Machine translation now rivals average human translation for language pairs where large,
handtranslated datasets are available. According to Google research, which uses humans to
score machine translations on a scale from zero (complete nonsense) to six (perfect), in 2015
Google Translate received a grade of 3.6—far worse than the average human translator who
receives scores like 5.1; by 2016, Google Translate had hit numbers like 5.

Today, machine translation is on smartphones, laptops, and tablets. Free apps like Google
Translate and iTranslate Voice are now quite good across the major language pairs, and
machine translation has been widely adopted. Google carries out a billion translations a day
for online users. YouTube has instant machine translation for many foreign-language
YouTube videos, showing instantaneous results in the form of English subtitles. Instant and
free-spoken translation is also possible with the Skype Translator add-on option.

Computing power and massive new datasets are the reasons why machine translation has
become so good, so fast. Machine learning trained AI-systems can now recognize language
patterns well enough to carry out human-level translation, with the real constraint being the
lack of human-translated sentences. For language pairs where the data is available, translation
is good. For others it is poor.

The switch came in 2016 when Jeff Dean switched Google Translate AI team from hands-on
programming to machine learning. The data constraint was relaxed when the UN posted
online a dataset with nearly 800,000 documents that had been manually translated into the six
official UN languages: Arabic, English, Spanish, French, Russian, and Chinese. The EU has
also released huge datasets along with the EU Parliament. The Canadian parliament has also
followed suit. With data and the requisite computer power to process it, the Google
Translation app improved more in a month than it had in the previous four years.

Machine translation will fundamentally alter the global supply of service workers. About 400
million people speak English as their first language, and if one includes proficient
secondlanguage speakers, there are something like a billion people who could sell services in
English online. It would seem that machine translation might multiply this number by two or
three, creating a tidal wave of online talent.

6.2 Work reorganization and telecoms

Another factor that is accelerating the trend toward remote work is the way in which US and
European companies are reorganizing themselves to facilitate slotting in telecommuting
workers. These companies are now using new collaborative platforms such as Business
Skype, Slack, Trello, Basecamp, and others that help organize communication among team
members.

These new collaborative platforms are designed to facilitate all manner of team
communication, ranging from text chats, emails, and discussion groups to phone calls,
Facebook posts, and multi-person video calls with screen sharing. Technology has also
facilitated telemigration via better communication.

Telecommunications are an essential ingredient in the globotics transformation and have been
improving in line with the explosive pace in the improvement of digital technology. This
started out with telephone calls becoming cheaper. Later, mobile phones became universal.
And once cheap and widely-available broadband had become more reliable, Skype and other

8
videoenabled communication technologies further stimulated the explosion. But in recent
years, things have gone even further.

Recently, new technologies have been creating more options between talking on the phone
and a face-to-face meeting. These new options are a long way from perfect, they may never
replace physical human contact, but they nevertheless present far cheaper and faster options
than physical meetings, and at the same time deliver much higher quality than standard phone
calls. The main new communication technology is called ‘telepresence’ and is already widely
used by big banks, consultancies, law firms, and governments. Telepresence makes it seem,
or almost seem, as if people are in the same place, even when they are not.

Augmented and virtual reality devices

One telecommunication technology that is advancing fast uses augmented reality or the
projection of a digital image onto reality via a headset, glasses or even a smartphone screen.
The two key supporting technologies are augmented reality (AR) and virtual reality (VR).
Many companies, both start-ups and giants like IBM, are using AR and VR to improve
remote collaboration. They are redefining what it means to work side-by-side with someone
and are going a long way toward taking the ‘remote’ out of remote work.

The big selling point of AR is that it allows an expert sitting somewhere else to ‘augment’ the
reality you are looking at through a video screen on your phone, tablet, or laptop. They can
explain what you need to do almost as if they were standing by your side. They do this by
placing computer graphics on your screen in a way that looks like it is part of the reality you
are looking at. Instead of ‘talk you through it’, they show you with arrows, circles, and the
like. There is no need for the remote expert to ‘paint a word picture’ of what needs to be done
since both workers are looking at the same reality on the screen, augmented by things added
by the expert.

Virtual reality is a far more immersive experience than AR. It completely hijacks your visual
and audio channels, replacing them with a computer-generated reality. It is a little
disorienting since you have no direct connection with where you are actually sitting. To date,
the images are too grainy to fully convey micro-expressions and the like, but body language
has amazing effects on how you perceive people.

There are other forms of telecommunication technology in early testing stages such as
‘holographic telepresence’. This projects real-time, three-dimensional images of people
(along with audio) in a way that makes it seem as if the remote person is right next to you.
This is the stuff of science fiction, but it is not unimaginable—having been used in the 2017
French presidential election and the 2014 Indian election.

7. What globotics and telemigration means for development strategies:

Conceptualization

If the automation of manufacturing means localized, jobless manufacturing, then many

national development strategies will need rethinking. The changes may not appear for 10 or
20 years, but this is in line with the timeline of most national industrialization strategies, and
so looking ahead is of vital importance.

9
The changing nature of manufacturing is not a new point. ‘A nascent yet growing body of
evidence has begun to challenge the long-held tenets of economic development that
industrialization is the prime engine of growth’, write Loungani et al. (2017). This message is
also clear in Hallward-Driemeier and Nayyar (2017): ‘Globalization and new technologies
are impacting the desirability and feasibility of what has historically been the most successful
development strategy’, namely manufacturing. ‘Many of the pro-development characteristics
traditionally associated with manufacturing—tradability, scale, innovation, learning-by-doing
— are increasingly features of services.’ As Schwarzer and Stephenson (2019) put it: “While
industrial development has played a key role in export-led development trajectories in the
past, … ICT-enabled services in particular offer potential for export diversification that defy
the logic of traditional paradigms by relying purely on electronic cross-border delivery,
making it accessible even to countries with underdeveloped physical trade infrastructure.”

This section explores conceptualization issues and asks the question: how may switching to
service-led development change thinking about national development strategies? Before
turning to the economic logic of a service-led development strategy, we review why
manufacturing was the linchpin of so many countries’ development strategies.

7.1 Why industry was viewed as so important to development

Most of today’s rich nations became rich by industrializing. Since 1970 or so, the same
nations have been deindustrializing due to globotics, but their historical growth take-offs
were closely associated with a rapid structural transformation that shifted workers from farms
to factories. The intense industrialization-growth correlation continued in the post-war period
with the four
‘tigers’ (Hong Kong, Korea, Singapore, and Taiwan) and a few second-generation tigers such
as Thailand, Vietnam, the Philippines, Malaysia, and Indonesia. For these nations,
industrialization was closely linked to export-led manufacturing, not just manufacturing.

This plain-as-day evidence is explained by a number of well-known mechanisms that made

manufacturing-led development particularly attractive for developing-nation governments.
First, rising manufacturing absorbed a lot of unskilled workers with minimal formal
education, especially in unskilled-labor-intensive sectors such as clothing and footwear. They
could, as it were, walk off the farm and into factories. In countries with growing masses of
relatively uneducated young people and few jobs outside of subsistence agriculture,
manufacturing was a blessing for social and political stability.

Second, manufacturing activities were development ‘escalators’. Marked by scale economies

and technology spillovers, each extra manufacturing job was viewed as helping the nation
and not just the individual worker. At least since the massive opening embraced by
developing nations as part of the second unbundling, these manufacturing jobs were tied to
flows of international trade, investment and know-how as rich-nation firms offshored
production (and manufacturing know-how) to nearby developing nations (Baldwin 2014).

This knowledge transfer aspect was especially powerful in East Asia where proximity of
technology leaders like Japan, Korea, Taiwan, Hong Kong and Singapore fostered such
offshoring. Today, for example, about 40% of world manufacturing is done in countries that
are touched by the Beijing, Tokyo and Hong Kong triangle. This region is quite simply one
of the most attractive locations for offshoring stages of manufacturing. The mechanisms—
mostly involving global value chains linking trade, investment, manufacturing, and
productivity growth included alleviation of supply bottlenecks, elimination of small-market

10
demand constraints, transfer of know-how, and connection to worldwide sales networks
(Taglioni and Winkler 2016). As an empirical matter, productivity in developing-country
manufacturing seems to converge toward that of rich nations with the global technological
frontier, regardless of policy and institutional determinants (Rodrik 2013). Development
theorists wove these facts into elegant intellectual frameworks.

7.2 Manufacturing-led development theory

The conceptualizations behind manufacturing-led development have a distinguished pedigree

starting with the glory days of what Krugman called ‘high development theory’, namely from
Rosenstein-Rodan (1943) to Hirschman (1958). We should probably add in the contributions
from New Economic Geography as exemplified by Krugman and Venables (1995).

In these frameworks, development is viewed as a virtuous cycle driven by external

economies, with each turn of the wheel providing economic momentum for the next turn. The
selfreinforcement came from an interaction between economies of scale at the level of the
individual producer and the size of the market. The concept came in two main fashions: ‘Big
Push’ and ‘Leading Sectors’.

Paul Rosenstein-Rodan argued for an economy-wide ‘Big Push’ effort by governments to

break out of underdevelopment. Productivity and incomes were higher in industry than
agriculture, but poor-nation firms failed to industrialize spontaneously because they could not
attain the minimum efficient scale in the face of lackluster local demand. The lack of local
demand, however, was due to the low incomes stemming from the lack of industry.

Albert Hirschman advocated instead for pushing for growth in particular sectors which would
— via ‘backward’ and ‘forward’ linkages—ignite growth in the rest of the economy. As
theories go, these were just about perfect in terms of shaping the thinking of policymakers
around the world. They were simple but not simplistic. And they were optimistic. Any nation
could do it. But as is often the case, ideas were the easy part. Implementation was the hard
part.

A key problem was the demand-creation part, regardless of whether it was to be created by an
economy-wide push or sector-specific policies. In the initial post-war decades, import
protection was the go-to policy. ‘All present-day industrial and developing countries
protected their incipient manufacturing industries producing for the domestic market’, wrote
Bela Balassa in his 1981 book, The Newly Industrializing Countries in the World Economy.

In this so-called import-substitution industrialization (ISI) strategy, importing was necessary

for key inputs, but the output was headed for the local market. Early thinkers like Raul
Prebisch, touting ‘export pessimism’, doubted that a developing nation could rely on export
to gin up the necessary demand for the output of their factories, especially since many
Northern markets were still heavily protected. While a very small number of developing
nations managed to trigger what looked like a self-fueled virtuous cycle in manufacturing
(e.g., Korea and Taiwan), ISI failed in most places. More precisely, it created jobs in labor-
intensive sectors like footwear, clothing, and furniture, but the step to starting competitive
heavy industries proved too much for the dozens of nations who tried it. The 1980s debt
crises drew a line under ISI.

Attention then turned to a new twist of manufacturing-led development—the ‘Washington

Consensus’. This leaned on Hirschman-Rosenstein-Rodan conceptualizations of the sale-

11
scales conundrum (need scale for sales, but need sales for scale), but sought to overcome it
with different tactics. Governments and markets would do the job, and exports were critical
to achieving scale. Import-competition took on a new positive role as a way to avoid abuse of
dominant positions in the domestic market. This too passed. By 2002, few of those that tried
it managed to launch their industry on an upward trajectory.

7.3 A stylized ‘old-school’ development journey

To set the stage for a consideration of service-led development, it may be useful to present a
stylized version of development challenges and stages that manufacturing-led development
strategies were meant to address in the Hirschman-Rosenstein-Rodan-Prebisch
conceptualization.

The externalities and spillovers critical to these accounts are economically beneficial for
lowincome countries who can seize them. But the same features make industry ‘lumpy’.
That, in turn, makes it hard to start industrialization and keep it going. The basic notion is
that a country could be good at industry if only it had more industry. It is a classic multiple
equilibrium problem. A nation with an industrial base can be globally competitive in a wide
range of final goods. The competitiveness, in turn, provides the sales necessary to justify that
industrial base. A classic manufacturing-export-led development story involves a shift from
the agrarian to the industrial equilibrium and all that this entails. Investment in capital and
skills booms, exports rise, foreign investment grows, infrastructure is constructed since it is
regarded as a worthwhile investment, people migrate from farms toward urban areas, and
young people find education attractive, etc. Here we depict the canonical development
journey graphically to spotlight the key role of external economies of scale.

Colonialism and underdevelopment

Our stylized development journey opens in the colonial period when the nation under study
either has never had industry or was deindustrialized during colonialism. That is, we start
with the nation completely specialized in agriculture goods in the sense that all labor in the
trade sectors (manufacturing and agriculture) are in the agrarian sector. This is point E0 in the
diagram (Figure 9).

There is a nontrade service sector, which employs LS workers, but this doesn’t yet enter the
analysis. The labor needed to provide the services demanded at the equilibrium price (which
is LS

in the diagram) are just subtracted from the nation’s labor endowment, 𝐿𝐿, and the remainder
is divided between the traded sectors, agriculture, and manufacturing. For simplicity,
manufactured and agricultural goods are freely traded under colonialism.

Given the external economies in manufacturing and the assumed constant returns in
agriculture, industry is marked by an upward sloped value of marginal product for labor in
manufacturing (VMPLM) and a flat value of marginal product for labor in agriculture
(VMPLA). The VMPL curves are equal to the price (which is fixed by free trade) and a
physical marginal product of labor, MPL. In manufacturing, it is upward sloped due to
external economies of scale (i.e., the marginal cost of production falls with the economy-

12
wide level of production, but individual firms perceive constant returns with respect to their
own output). In agriculture, the VMPL is flat due to assumed constant returns.1

Figure 9: Schematic diagram of manufacturing-led development

Euros

VMPLMNT
VMPLMFT
E’
h w=VMPLAFT

E1
𝐼𝐼𝐼𝐼
𝐿𝐵𝐵 𝐿𝐵𝐵𝑀 𝐿𝐵𝐵 labour
E0 𝐿𝐼𝐼𝑀 𝑀 𝑀
LS 𝐿𝑀
𝐵𝐵
𝐿 𝑀
𝐵𝐵
𝐿𝑀
𝐿𝑀

Source: Authors’ elaboration.

This setup has two stable equilibriums. In one equilibrium, all workers are in industry, so the
VMPLM is higher than VMPLA and thus the wages are higher in manufacturing (point E’).
This is a stable outcome since the higher wage keeps all the workers in manufacturing. In the
other equilibrium (point E0), no one works in manufacturing, so the wages in manufacturing
are lower than they are in agriculture. This too is stable.

A critical point for the analysis is the level of employment in manufacturing, where the two
VMPLs intersect (𝐿𝐿𝐵𝐵𝑀𝑀 in the diagram). Here, the ‘B’ stands for ‘breakpoint’ since if
somehow LM arrived above the breakpoint, industrialization would be self-sustaining (as
shown by the rightpointing arrows). Any change in LM that pushes it to a point to the right of
the break point will, eventually, result in workers leaving industry for agriculture (as shown
by the left-pointing arrows). The breakpoint, 𝐿𝐿𝐵𝐵𝑀𝑀, might also be called the ‘minimum
efficient scale’.

To think about this, note that if VMPLM were above w, the country would be price-
competitive in the world market for manufactured goods when paying the wage, w. More
specifically, if firms paid w for labor but had MPL corresponding to the VMPLM above the
line at world prices, the firms could break even by charging a price below the world price.

Independence and import substitution

In the next phase of the development journey, the nation gains independence and shuts out
foreign manufactured goods in an effort to get the virtuous cycle spinning. The VMPLM now
depends on the domestic price and the domestic price falls as output rises. Assuming the right

1 Making agriculture subject to diminishing returns is a simple but unenlightening extension, as long as the
model displays multiple equilibriums in manufacturing.

13
combination of demand and scale elasticities, the new curve for industry is VMPLMNT and it
is downward sloped (NT stands for nontraded). The result is that some labor moves into
manufacturing, namely, it rises to 𝐿𝐿𝐼𝐼𝐼𝐼𝐼𝐼𝑀𝑀 where the superscript stands for ‘import
substitution industrialization’.

The critical issue for manufacturing-led development is whether 𝐿𝐿𝐼𝐼𝐼𝐼𝐼𝐼𝑀𝑀 is above or below
the breakpoint level, 𝐿𝐿𝐵𝐵𝑀𝑀. If it is above, then import substitution will produce a self-
sustaining industrialization with an endpoint of E’ (all labor in manufacturing). If it is below
the breakpoint, the ISI strategy will lead to stagnation. If the nation liberalizes imports, it will
deindustrialize back to E0 since with open trade, the upward sloped VMPLM is the relevant
one.

One strategy would be to try a ‘Big Push’ into industry, with the government pushing or
pulling labor into manufacturing. If the effort isn’t big enough, and let’s say the effort takes
the economy to point 𝐿𝐿𝐵𝐵𝐵𝐵𝑀𝑀 where the superscript stands for ‘big push’, then the economy
becomes stuck. And all efforts to expand industry will result in losses. If the push takes the
economy beyond the breakpoint, the ‘Big Push’ is successful and industrial development
becomes selfsustaining. Clearly a large home market is helpful here.

Washington Consensus and export-led industrialization

For many nations, the ISI and BP efforts failed to spark endogenous industrialization. In the
next phase, our template developing nation tries the Washington Consensus. This involves
improving institutions and lowering barriers to trade and investment.

Improved institutions are reflected in two ways in the diagram (Figure 10). On the
demandenhancement side, it shifts out the VMPLMNT curve since the better institutions it
effectively increases the size of the domestic market (less waste, better contracting, etc.).
What this does is raise 𝐿𝐿𝐼𝐼𝐼𝐼𝐼𝐼𝑀𝑀 . On the supply side, it rotates the VMPLMFT line counter-
clockwise. This is due, on one hand, to improvements in production efficiency that come
with imported know-how, etc. This raises the MPL part. On the other hand, openness and
international integration also result in a higher net-of-costs price for domestic firms (this
affects the p part of VMPL). The penhancing aspects and MPL-enhancing aspects bring the
breakpoint, 𝐿𝐿𝐵𝐵𝑀𝑀, closer to 𝐿𝐿𝐼𝐼𝐼𝐼𝐼𝐼𝑀𝑀 . Figure 10: The effect of better institutions

14
Source: Authors’ elaboration.

As with ISI, these reforms may or may not be enough to get our template nation over the
hump and send it on its way to a pro-market outward-oriented take-off in exports,
industrialization, and economic growth. As history would have it, for a few countries, the
policy reforms got 𝐿𝐿𝐵𝐵𝑀𝑀 past 𝐿𝐿𝐼𝐼𝐼𝐼𝐼𝐼𝑀𝑀 , but for many, it didn’t.

If the group of ‘it was not enough’ nations applying the Washington Consensus is sufficiently
large, the Washington Consensus itself would be viewed as a failure by empirical economists
who are unfamiliar with multiple equilibrium economics. The next step would be to conclude
that no ‘big idea’ is working in development (Lindauer and Pritchett 2002).

Note that since the VMPLMFT depends upon the price received by domestic manufacturers
(rather than the price paid by customers abroad), remoteness plays an important role in the
position of the VMPLMFT curve. Countries that are remote from large markets, as measured
by, for example, market potential, would find it expensive to import essential materials,
intermediate inputs, and capital equipment. They would also find it difficult to acquire and
sustain relationships with their foreign customers. And finally, the distance would discourage
advancedeconomy firms viewing the faraway nation as a good place to locate offshored
stages of production.

In all phases of this stylized development journey, the key challenge to sparking
manufacturingled development was getting past the hump (the point 𝐿𝐿𝐵𝐵𝑀𝑀). Beyond this
point, industrialization became a self-driving mechanism. Before it, nothing worked. Since
industry was marked by scale economies, it was very difficult, and usually impossible, for all
but the largest nations or those with excellent ‘market potential’ due to their geographical
location, to get over the hump with import-substitution policies, ‘Big Push’ policies, or
Washington Consensus policies.

The source of this failure might be hard to detect with the usual empirical approaches. If one
runs linear regressions of ISI and Washington Consensus policies on development outcomes
— treating all nations as random draws from a single distribution—the results may be
confusing. The policies could all be ‘working’ in the sense of narrowing the gap between the
employment level and the breakpoint, but not working in the sense of triggering
industrialization. As is wellknown, standard econometrics has trouble estimating threshold
effects without really big data sets. Since so few developing nations have managed to
industrialize rapidly, such data is unattainable.

In a nutshell, the key determinants in our recounting of this development journey are
domestic market size, institutions, and remoteness. For small remote nations with bad
institutions, neither ISI nor the Washington Consensus would work. For big favorably-
located nations, both development strategies could work.

The service sector was left completely in the background since that was where it was put in
most traditional thinking on national development strategies.

The traditional (minor) role of services in development

Development economics has long relegated services to the side-lines or ignored them
altogether.

15
Loungani and Mishra (2014) call it a deeply rooted prejudice against the service sector.
Adam
Smith in his famous tome cast aspersion on the social value of service provided such as
‘churchmen, lawyers, physicians, men of letters of all kinds, players, buffoons, musicians,
operasingers, opera-dancers, etc.’ (cited in Loungani and Mishra 2014). Karl Marx
considered many types of services as ‘faux frais’ of production—activities that were incurred
in the productive use of capital but which do not themselves create any value-added. This
was picked up on Soviet planning where services were downplayed compared to heavy
industry—and it was the success of this planning up to the late 1950s that inspired many of
the early post-war development thinkers. Similarly, Baumol (1967) fostered the view that
services are a sector resistant to improvements in productivity. The same message came
through in Ghani (2010)—a whole book devoted to illuminating the service-led growth in
South Asia.

More formally, Eichengreen and Gupta (2013) noted that the structural transformation
literature, which is intimately tied to that of development strategies, downplayed the role of
services:

The pioneers of the literature on structural change, such as Fisher (1939) and Clark (1940),
emphasized the shift from agriculture to industry in the course of economic growth and they
in fact said little about the share of services. Kuznets (1953) concluded that the share of
services in national product did not vary significantly with per capita income. Chenery and
Syrquin (1975) regressed the service-sector share of output on per capita income and per
capita income squared, concluding that the relationship was concave to the origin—that it
rose with per capita incomes but at a decelerating rate.

7.4 Can service exports spark a virtuous cycle?

Services played almost no role in the stylized development journey discussed above since
services were traditionally viewed as being marked by no scale economies, no positive
externalities, and few opportunities for export or inward foreign knowledge transfers.
Without these things, services cannot start a classic development spiral where an expanding
sector creates forces that encourage further expansion. But is this really true? Are services
really bereft of these features?

First, consider the possibility of exports and growth promotion. Service exports are, as we
showed above, growing, and growing faster than the export of goods. There is also some
evidence that exporting more sophisticated services is pro-growth (Mishra et al. 2011).

What about Baumol’s disease? The lack of productivity growth in services, or as Baumol
phrased it: one still needs four people to play a quartet despite centuries of technological
progress. That too turns out to be untrue, at least in its baldest form. In the US, Triplett and
Bosworth (2004) estimated that services accounted for over 70% of labor productivity growth
in the New Economy boom of the late 1990s. Ghani (2010) documents that after 2000, labor
productivity in India, Pakistan, and Sri Lanka grew faster in services than in manufacturing.

Figure 11 shows a decomposition of growth by sector in India and China. In China, factor
productivity was a big driver in manufacturing (as would happen in the case of external
economies and self-sustaining industrialization), but not so much in services. In India, the
order is reversed, even if India’s growth was slower overall. At the least, this suggests that
services, like manufacturing, can be a source of progress.

16
Some services also seem subject to external economies of scale. Inside nations, the notion
that service sectors are subject to agglomeration economies or external economies of scale is
widely taken as received wisdom, at least by urban economists. In advanced-economy cities,
the existence of a substantial wage premium is well documented (Combes and Gobillon
2015).2 These sorts of premiums are the hallmark of external economies as they indicate that
the whole is more than the sum of the parts. More to the point, in our development journey
story, the whole of getting more workers in manufacturing was to be able to sustain higher
wages. In terms of theory, authors such as Duranton and Venables (2018) draw an upward
sloped VMPL for cities that looks exactly like the VMPLMFT in the diagrams.

Figure 11: Sources of growth, India and China, 1993–2004

Source: Based on data from Bosworth and Collins (2008).

The assumption that services are subject to something akin to external economies is indeed
the heart of urban economics (Black and Henderson 1999) and the facts behind this
assumption are quite undeniable. The economies of the richest cities in the world are based
almost entirely on services and often service exports. The people in the service sectors earn
wages far beyond the average. Moreover, the cities themselves seem to be marked by
external economies of scale judging from the high price of apartments in city centers. The
good jobs are in big-city service sectors since cities are where the talented people are. As Ed
Glaeser puts it, smart people move to cities and make each other smarter. In terms of explicit
modeling, Robert-Nicoud (2008) is a clear example of incorporating trade in services into a
standard New Economic Geography model.

Overall, national development strategies in the digital era may do well to look to urban
economics and New Economic Geography for inspiration and guidance.

2 Combes and Gobillon (2015) review the empirical literature and find the elasticity of wages with respect to
city population is typically around 8–10% but much of that relationship is driven by the sorting of more skilled
individuals into larger cities. Correcting for these biases leads to smaller elasticities of 2–5%. It is also the case
that workers may learn more over time in larger cities.

17
8. Globotics and development mindsets

The hardest aspects of development are only marginally changed by globotics for the very
simple reason that the trade-and-development component of development is not the hardest
part. Even in quite open economies, most economic activity is by and for local citizens.
Getting it to work requires all sorts of difficult things like good roads and ports, good
institutions, good education, good healthcare, trust among citizens, trust by citizens of
government, and much more.

But the globotics transformation is likely to radically change the way we think about
development—if the posited thought experiment comes true. If labor-cost-based trade in
manufactured goods comes to an end and services become freely traded, ways of thinking
about development will have to change. This is not a novel thought.

The Pathways for Prosperity Commission’s 2018 report Charting Pathways for Inclusive
Growth lists ‘Global trade in services’ as its Pathway Three. The report lists the main ways to
unlock the pathways, the most relevant of which is about how governments and businesses
can create a digital-ready country. Many of the recommendations are akin to those suggested
by UNCTAD in its many publications on e-readiness that stress five pillars: enabling digital
infrastructure, enabling legal and regulatory frameworks, enabling human capital, enabling
finance, and enabling coordination. Mattoo (2018) discusses what the new emphasis on
services means for international cooperation efforts, and Heuser et al (2017) look at the role
of services in global value chains.

A study that focuses on creating digitally-enabled jobs for African youths (Mastercard
Foundation 2019) suggests a few ‘no regret’ measures that policymakers could take. These
include the collection of better data locally, close monitoring of international developments,
the provision of training for local policymakers on digital economy matters, promoting the
provision of digital ‘soft’ commerce skills (such as digital marketing and relationship
management) as well as hard skills (such as coding), and embracing a ‘test-and-learn
approach’ to deal with the uncertainties and rapid pace of change. The 2016 ECLAC report
Innovation and Internationalization of Latin American Services presents many ideas for how
governments should think about and prepare for digitech’s impact on development
(Hernández et al. 2016).

Here we do not repeat or even catalogue the suggestions. Rather, we attempt to focus
attention on how services are different when it comes to development mindsets.

8.1 How are services different?

There are two critical shifts in thinking when it comes to service-led development that we can
summarize in two pity aphorisms.

Stop thinking factories, capital equipment and technology. Start thinking cities,
people and training.

The key to industrialization was the ability to have the right equipment and technology in
place and to line up the factories with sufficiently large customer bases. In the traditional
development strategy, labor in the manufacturing sector is not viewed as a relevant
constraint. It is a bit of a sideshow since one of the great features of manufacturing is the

18
ability of workers to practically walk out of the rice fields into a factory and to start being
productive with very little preparation. This scenario is not true of modern services.

When it comes to modern services, the people are the ‘capital equipment’, so to speak. And
people do not come with embedded technology, unlike a robot welder from Germany.
Foreign know-how may be important, but for many types of export services, for example
coding, copyediting, or project management, the technology represents a sideshow. It is the
skills and experiences of the people, the service providers, that are the real constraint.

Joining service value-added chains require less of a great push than the development of an
industrial base. However, the accumulation of human capital may take a longer amount of
time, compared to the accumulation of physical capital.

Secondly,

Stop thinking factories and start thinking of cities as productive platforms.

Cities are where people meet and form local networks for face-to-face connections, where
people exchange ideas, and competition among ideas plays out. Cities facilitate matching
between service workers and service firms. As the economic geographer Enrico Moretti puts
it, cities become ‘brain hubs’ (Moretti 2013). Workers and firms implicitly benefit from each
other’s knowledge creation via face-to-face interaction and social networking. This point is
not new.

In 2010, the Netherlands Bureau for Economic Policy Analysis wrote a study of how the
Netherlands could future-proof its economy and the answer was: cities. ‘At the beginning of
the twentieth century, manufacturing firms settled near each other in order to benefit from
knowledge spillovers in the development of electricity. Cities should not be thought of as
mere collections of people, but rather as complex workspaces that generate new ideas and
new ways of doing things’ (Bas ter Weel et a. 2010).

Governments should start thinking of cities as production hubs, not just living quarters. Cities
should be conceptualized as geographic centers for face-to-face interactions that foster the
production of export-oriented services. Winning cities will attract high-quality jobs and lock
them in with agglomeration forces.

Another change in thinking is actually a continuation of the thinking about manufacturing.

We are all by now familiar with the notion of the role of the second unbundling in
manufacturingled development. The idea that a developing nation can join a value chain is
clear. The same can be held true in services.

9. Concluding remarks

This paper seeks to think through some of the implications that digital technology may have
for developing nations, for their development strategies and for the broader world. Our
conclusion is that the service-led development path, such as the path that India took, may
actually become the norm rather than the exception. Since success in the service sector is
based on quite different factors than success in manufacturing, our conclusion suggests that
development strategies and mindsets will have to change. While change is always hard, this
is a fundamentally optimistic conclusion for developing nations for a very simple reason.

19
Digitech will allow many emerging markets to directly export the source of their comparative
advantage (i.e. labor which is low cost given its productivity) without having first to make
goods with that labor and then export the goods. One way of thinking about comparative
advantage trade in a Ricardian model is that trade in goods is a veil for trade in labor services.
Digital technology is merely pulling back the veil. The resulting expansion in service trade is
likely to be an overall net export gain for emerging markets and an overall net import gain for
developed economies.

20