## wp18223

## Source details

**Canonical URL:** [wp18223](https://www.imf.org/-/media/files/publications/wp/2018/wp18223.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18223.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18223.pdf.json)

---

### 2.1 Drivers of Structural Transformation and Industrialization
- Definition: Structural transformation = secular shifts in sectoral decomposition of employment and output between agriculture, manufacturing and services that accompany increases in GDP per capita.
- Typical historical pattern: agriculture → manufacturing → services; held "pretty well until the early 1970s/80s" but recently challenged by globalization and technological revolution.
- Key contemporary phenomena motivating revision of theory: deindustrialization and servicification; declining share of manufacturing production and jobs "over the past two decades."
- Theoretical drivers:
  - Push factors (agriculture-origin):
    - Lewis (1954): surplus labor in agrarian economies; freeing this labor for manufacturing can jump start development.
    - Schultz (1953): low productivity in agriculture is a technological feature; agricultural improvements (e.g., green revolution beginning in the 1960s) required to trigger development.
    - Both rely on non-homothetic preferences with highly inelastic demand for agricultural goods.
  - Pull factors (industry/services-origin):
    - Productivity gains in industry and higher urban wages entice labor out of agriculture.
    - Public policy can support pull factors by promoting manufacturing and urban-centered activities.
- Mechanisms in models:
  - Non-homothetic preferences + productivity changes explain labor movements: early stage non-homotheticity moves labor out of agriculture; later industrial productivity growth moves labor from industry to services.
  - Swi̧ecki (2017): three-sector GE model—industry productivity increases + non-homothetic preferences account for agrarian → industrial → post-industrial path.
  - Consumption specification: income elasticities of agricultural goods < 1 and services > 1 generate empirically consistent consumption bundles with income expansion.
- Fundamentals and complementarities:
  - Private fundamentals: capital, technology, skills, innovation—complements to labor for transformation.
  - Public fundamentals: infrastructure and institutions to enable markets and coordination.
  - Coordination failures and network effects can inhibit transformation even when private fundamentals improve.
- Historical/policy lessons:
  - Rostow (1960): sufficient investment (domestic or foreign saving) important for "take-off".
  - Rosenstein-Rodan (1943): "big push"—public policy can loosen multiple constraints and exploit economies of scale.
  - Policy experiences:
    - ISI: supported nascent manufacturing (Brazil, India) but reduced competitiveness long-run; India experienced License-Permit Raj until early 1990s.
    - EPI: smaller East Asian economies (South Korea, Taiwan) used export promotion with government support to sustain industrial productivity and transition to services.
- Growth diagnostics (Hausmann, Rodrik, Velasco, 2007):
  - Decision-tree approach to identify binding constraints; target one or a few constraints to improve efficacy and limit distortions.
  - Distinguishes issues like low appropriability versus high cost of finance to guide policy choice.
- Preserved timestamps and references: green revolution beginning in the 1960s; Lewis (1954); Schultz (1953); Rosenstein-Rodan (1943); Rostow (1960); Hausmann, Rodrik, and Velasco (2007); Swi̧ecki (2017); Rodrik (2012, 2015); World Economic Outlook, April, 2018.

### 3.1 Opportunities and Challenges on the Demand Side
- Openness and specialization:
  - Open economies can specialize and exploit economies of scale in manufactured goods, enhancing productivity.
  - Closed economies need diversified domestic industrial bases; small populations limit economies of scale.
  - Open goods trade subjects countries to comparative advantage outcomes (favorable or unfavorable).
- Empirical regularity (Felipe and Mehta, 2016):
  - Global share of manufacturing in employment ~14% (1970–2010).
  - Global share of manufacturing in output ~16%-17% (1970–2010).
- Implications:
  - Countries with comparative advantage in manufacturing can capture larger shares of global demand and accelerate structural transformation via private fundamental upgrading (technology, skills).
  - South Korea and Taiwan cited as successful examples.
- Reallocation scenarios for manufacturing:
  - Most favorable: developed countries’ rising opportunity costs of producing manufactures shift production to industrializing developing countries (consistent with Vernon (1966) product cycle).
  - Less favorable: many low-income countries simultaneously pursue transformation but not all can benefit due to lack of comparative advantage (example: Brazil).
  - Least favorable: country lacks comparative advantage, becomes net importer of manufactures, finances imports by exporting agricultural goods or natural resources, failing to industrialize (examples cited: Ghana, Malawi, Botswana in SSA).

### 4.2 Premature Deindustrialization: A Summary
- Global patterns (Felipe and Mehta, 2016):
  - Global manufacturing employment share remained ≈14 percent.
  - Global manufacturing output share remained ≈17 percent.
  - Average manufacturing productivity mirrors global productivity across sectors.
  - Stability from 1970 to 2015 implies intense competition for a stable share of world manufactured output.
- Rodrik (2015) on premature deindustrialization:
  - Countries that began deindustrializing since 1990: peak manufacturing employment share = 18.9 percent at per capita income level = 4,273 constant 1990 dollars.
  - Countries deindustrializing prior to 1990: peak manufacturing employment share = 21.5 percent at per capita income level = 11,048 constant 1990 dollars.
  - Premature deindustrialization → lower manufacturing employment shares and earlier income peaks → retarded productivity growth as labor moves into lower-productivity agriculture and services.
- Country case studies:
  - Ghana:
    - Annual growth ≈5% over 2001-2010.
    - Structural break beginning in 1992: intensified labor movement from agriculture to services, sharp reduction in manufacturing.
    - Manufacturing share fell to 5.14% in 2014, slight recovery to 5.33% in 2015.
    - Most labor moved into low- and average-productivity services (productivity of one-tenth and one-half of high-productivity services).
  - Nigeria:
    - Growth ≈6.8% over the last decade (Ajakaiye et al., 2016).
    - Oil sector accounts for 20%-30% of GDP, 70% of government revenues, and 85% of exports.
    - Services share rose from 18% in 1970 to 44% in 2014.
    - Manufacturing share fell from 12% to 6% over same period.
    - Characterized as transforming directly to a tertiary service economy without intermediate industrialization.
  - Botswana:
    - GDP grew at 12 percent per year from 1960 to 1990; joined upper-middle-income status in 2005.
    - Manufacturing share of GDP rose from 1.4% in 1960 to 6.6% in 2010.
    - Services share rose from 8.5% to 50.6% of GDP over same period.
    - Share of wholesale and retail trade in GDP rose by 18.2 percentage points.
    - Performance linked to landlocked geography, integration with South Africa, and SACU-led trade liberalization.
- Causes of premature deindustrialization:
  - Global constraint: stable global manufacturing shares limit opportunities for many countries to expand manufacturing via world demand.
  - Public fundamentals and complementarities: foreign capital/technology/skills require infrastructure and institutions; poor infrastructure deters investment (e.g., many SSA countries).
  - Globalization and supply-chain reallocations: capital and technology move to countries with stronger public fundamentals; component-production integration benefits some countries without full vertical integration.
  - Excluded countries become importers of manufactures, terminating structural transformation early; labor remains in agriculture or moves to lower-productivity services.
- Productivity and growth consequences:
  - Rodrik (2012): average rate of convergence in manufacturing industries = 2.9 percent per year; translates into additional boost to productivity of over 6 percent for industries in bottom 20 percentile of his sample.
  - Premature deindustrialization blocks employment absorption into manufacturing "escalator" industries, slowing convergence and leaving per capita income on lower growth trajectories; in some cases near stagnation.
  - Rise of labor-saving technology (e.g., AI-enhanced robotics) may further complicate late industrialization.
- Middle-income trap and links to deindustrialization:
  - Examples: Brazil, Malaysia, Mexico, Peru failed to move steadily up the value chain.
  - Trap seen as a "bad, but stable equilibrium" requiring external shock or significant policy to escape.
  - Han and Wei (2016): heterogeneity in trajectories; LICs take longer to attain middle-income than middle-income to attain high-income status.
  - TFP slowdowns observed in 1980-1990 decade for Brazil, Argentina, Peru, Mexico; ISI policies contributed by limiting quality/competitiveness.
- Policy-relevant implications:
  - Conditions for industrialization: strong public fundamentals (infrastructure, institutions), openness, and complementarities to attract foreign capital, technology, and skills.
  - Medium- to long-term opportunity: if large economies (e.g., China) transition to post-industrial services, space could open for smaller countries to fill manufacturing roles—contingent on other countries escaping middle-income malaise and recipient countries having sufficient public fundamentals.
  - Policy takeaway: latecomers must prioritize public fundamentals, infrastructure, institutional quality, openness, and policies promoting complementarities to capture manufacturing opportunities and avoid premature termination of structural transformation.

### 5.2 Can Services Do the Trick?
- Patterns in latecomers (Africa):
  - Countries with labor movement into traditional (nontradable) services: Ethiopia, Ghana, Kenya, Malawi, Senegal, Tanzania, Zambia.
  - Wholesale and retail trade sector saw largest employment share increases over last two decades (1990-2000 and 2000-2010) in almost all listed countries (Enache, Ghani, O’Connell, 2016).
  - Formalization rate in wholesale/retail trade reported as 5.90% (Osei and Jedwab, 2017).
  - Movement into wholesale and retail not correlated across countries with sector productivity—observed both where sector productivity is low (Ghana, Kenya, Senegal) and high (Ethiopia, Malawi, Tanzania, Zambia) (Enache et al., 2016).
- Distribution costs:
  - Distribution costs >40% of retail price in the US and ≈60% of retail price in Argentina (Burstein, Neves, Rebelo, 2003).
- Assessment of traditional nontradable services:
  - May generate growth (unconditional cross-country convergence in services documented in World Economic Outlook, April, 2018), but unlikely to sustain long-term growth because they rarely generate sustained productivity increases.
  - A credible services-based alternative must rely on growth in high-productivity, tradable services.
- Indian experience (5.2.1):
  - Historical industrialization: 1950s–1960s rapid industrialization via ISI, focusing on domestic consumption and lower-quality manufactured goods.
  - Structural transformation in India relied more on agriculture → services than agriculture → manufacturing.
  - Sectoral shares (1960s to early 2000s):
    - Manufacturing employment share: from 9.8% to 12.4%.
    - Wholesale and retail trade employment share: from 4.88% to 8.27%.
    - Transport and storage employment share: from 2.06% to 3.32%.
    - High-productivity services (finance, insurance, real estate, business services) share: from 0.32% to 1.68% (five-fold increase).
    - High-productivity services share in 1980-1989 = 0.40% (increase mainly since 1990).
  - Macroeconomic outcomes:
    - India mitigated lack of industrialization and maintained secular per capita GDP increase over 40 years since first attaining GDP per capita of $800 U.S. (in 2000 dollars).
    - Per capita GDP growth: 1990–2015 = 4.81%; since 2000 = 5.63%.
  - Drivers of services-led growth:
    - Longstanding emphasis on scientific/technical knowledge and manpower since 1947.
    - Large pool of technically proficient English-fluent workers (legacy of British rule).
    - Large Indian diaspora in world IT centers providing networks (e.g., Y2K tailwinds).
    - Liberalization of information and communications sectors beginning mid-1980s.
    - IT sector development bypassed License-Permit Raj constraints on traditional industry.
    - IT revolution could bypass inferior physical infrastructure.
  - Limits:
    - Employment in high-productivity services increased by 1.28 percentage points.
    - Employment in other services (construction, wholesale and retail, transport and storage; excluding community and social services) increased by 4.91 percentage points.
    - Productivity in other services = one-third to one-fifth of high-productivity (communication and IT) counterparts.
    - Conclusion: IT/services growth substantial but ability to absorb large amounts of labor released from agriculture/manufacturing was limited.
- High-productivity services: potential and prerequisites (5.2.2):
  - Private fundamentals needed: appropriate technological knowhow and availability of technical manpower.
  - Public fundamentals and policies:
    - Minimum communication and IT infrastructure threshold required.
    - Government policies linking domestic activity to international markets (e.g., encouraging FDI) are generally desirable.
    - India’s diaspora networks reduced need for some government measures.
  - Trade-offs:
    - Public fundamentals: less onerous for services-led growth than for industrialization.
    - Private fundamentals: services-led growth requires stronger private fundamentals (technology and skills), further removed from unskilled labor than manufacturing.
  - Fundamental constraint: services-led growth often requires leapfrogging the production structure relative to domestic demand; depends heavily on global demand and exports.
- Need for longer-term policy perspective (5.3):
  - Traditional low-income focus: capital accumulation and infrastructure to strengthen private fundamentals.
  - Latecomers deprived of manufacturing-driven virtuous cycles risk being denied historical manufacturing opportunities.
  - Middle-income malaise problems (Agenor, 2016):
    1. inadequate infrastructure,
    2. diminishing marginal benefits from physical capital investments,
    3. exhaustion of cheap labor and imitation gains,
    4. insufficient human capital to compete in innovation,
    5. misallocation of talent to low-growth sectors,
    6. lack of access to funding for risky innovative investments,
    7. income inequality limiting educational access.
  - Policy implications:
    - Long-horizon focus: skills and innovation key challenges.
    - Public policy must be broad-based beyond infrastructure, FDI attraction, and business climate improvements.
    - Policies should address country-specific market failures and incentivize individual and business investment in skill upgradation and innovation to sustain long-run growth.
- Conclusion (section 6):
  - Manufacturing historically drove productivity growth and mass employment absorption but may no longer deliver similar outcomes for many low-income countries.
  - Many SSA countries experienced growth accelerations in the 1990s and 2000s without corresponding manufacturing growth.
  - Alternatives include improving agricultural productivity or transitioning to services (including rising non-tradable services in urbanizing "mega (consumption) cities").
  - Services-led growth viable for some countries with idiosyncratic favorable factors but unlikely to be universal; potential for broad structural transformation and large-scale labor absorption remains unclear.
  - Policy emphasis:
    - Manufacturing can absorb large amounts of labor and catalyze skill upgrades, technology improvements, and innovation.
    - In absence of strong private fundamentals, public policy must create incentives to strengthen technology, skills, and innovation.
    - Because private fundamentals evolve slowly, public policy should emphasize long-run, sustainable growth and structural transformation early.

*Source — IMF Working Paper (wp18223).*

### 2.1 Drivers of Structural Transformation and Industrialization......................................6

### 2.1 Drivers of Structural Transformation and Industrialization

### Introduction and framing
- Structural transformation: secular shifts in sectoral decomposition of employment and output between agriculture, manufacturing and services that accompany increases in GDP per capita.
- Typical pattern: agriculture (least productive) → manufacturing (normally most productive) → services.
- Historical observation: this textbook path "held pretty well until the early 1970s/80s" but has been "seriously challenged more recently" by globalization and technological revolution.
- Key contemporary phenomena challenging prior theories: deindustrialization and servicification.
- Motivation of the chapter: revisit policy and conceptual frameworks as pioneer theories of structural transformation are overturned by new phenomena.

### Theoretical drivers: push and pull factors
- Push factors (originating in agriculture):
  - Lewis (1954): surplus labor in agrarian economies (under-employment and disguised employment); freeing this labor for manufacturing can jump start development.
  - Schultz (1953): low productivity in agriculture is a technological feature; improvements within agriculture are required to trigger development (e.g., the green revolution beginning in the 1960s).
  - Both push theories assume non-homothetic preferences with highly inelastic demand for agricultural goods.
- Pull factors (originating in manufacturing and services):
  - Productivity gains in industry and higher urban wages entice labor out of agriculture.
  - Public policy can facilitate pull factors by supporting manufacturing and urban-centered activities.

### Mechanisms emphasized in models
- Non-homothetic preferences combined with productivity changes explain labor movements:
  - Early stage: non-homotheticity is key to move labor out of agriculture.
  - Later stage: continued industrial productivity growth drives movement from industry to services.
- Swi̧ecki (2017) three-sector general equilibrium model: productivity increases in industry plus non-homothetic preferences provide a "fully satisfactory account" of the path from agrarian → industrial → post-industrial economy.
- Consumption pattern implications: specification where income elasticities of agricultural goods < 1 and services > 1 generates empirically consistent consumption bundles as income expands.

### Role of private and public fundamentals
- Private fundamentals: capital, technology, skills, and innovation—necessary complements to labor for structural transformation.
- Public fundamentals: infrastructure and institutions that enable markets and coordination among private agents.
- Complementarities and coordination failures:
  - Structural transformation requires complex, interrelated changes in production, markets, and institutions.
  - Network effects and coordination failures among private agents can inhibit transformation even when private fundamentals improve.

### Historical and policy lessons
- Rostow (1960): sufficient investment (domestic or foreign saving) is important for the "take-off".
- Rosenstein-Rodan (1943): "big push"—public policy can simultaneously loosen multiple constraints, exploit economies of scale, and generate demand.
- Examples of policy outcomes:
  - Import-substituting industrialization (ISI): initially supportive for nascent manufacturing (Brazil, India), but long-run effects included reduced competitiveness; in India, compounded by License-Permit Raj until the early 1990s.
  - Export-promoting industrialization (EPI): smaller East Asian economies (South Korea, Taiwan) used EPI with government support to sustain industrial productivity increases and later transition to services.

### Growth diagnostics framework and policy targeting
- Hausmann, Rodrik, and Velasco (2007) growth diagnostics:
  - Provides a decision-tree approach (Figure 1) to identify the binding constraints at a point in time.
  - Objective: target one or a few binding constraints to improve efficacy and reduce distortions from policy interventions.
  - Example diagnostic distinction: low appropriability versus high cost of finance implies different policy responses.
- Implication: careful diagnosis helps direct policy effort to the most pressing constraints and generate growth benefits with least distortion.

### Summary of key conceptual points and timestamps preserved from the source
- Historical anchors: green revolution starting in the 1960s; the traditional path held until the early 1970s/80s.
- Important references and dates cited in the discussion: Lewis (1954); Schultz (1953); Rosenstein-Rodan (1943); Rostow (1960); Hausmann, Rodrik, and Velasco (2007); Swi̧ecki (2017); Rodrik (2012, 2015); World Economic Outlook, April, 2018.
- Noted phenomena motivating the revisit: declining share of manufacturing production and jobs "over the past two decades."

_Italic: Source — IMF Working Paper (chapter section titled "2.1 Drivers of Structural Transformation and Industrialization" as provided in the supplied PDF content)._

### 3.1 Opportunities and Challenges on the Demand Side

### 3.1 Opportunities and Challenges on the Demand Side

### Demand-side effects of openness and specialization
- An open economy allows specialization and fuller exploitation of economies of scale in production of manufactured goods, enhancing productivity.
- A closed economy would need to create a diversified industrial base to meet domestic demand, but small populations limit the ability to fully exploit economies of scale.
- Open goods trade exposes countries to the forces of comparative advantage, which can produce either favorable or unfavorable outcomes.

### Empirical regularity and implications
- As documented by Felipe and Mehta (2016), the global share of manufacturing in employment and output was remarkably constant over 1970-2010 at 14% (employment) and 16%-17% (output).
- Countries with a comparative advantage in manufacturing can capture proportionally higher shares of global demand and accelerate the first phase of structural transformation, benefiting from continued productivity growth in industry via upgrading of private fundamentals (technology and skills) through innovation and learning.
- South Korea and Taiwan are cited as examples of countries that leveraged global demand to accelerate industrialization.

### Scenarios for reallocation of manufacturing activity
- Most favorable scenario:
  - Rising opportunity costs of producing manufactures in developed countries reduce their comparative advantage.
  - Production shifts to developing countries undergoing structural transformation and industrialization (consistent with Vernon’s (1966) product cycle theory of international trade).
- Less favorable scenario:
  - A large number of low-income countries simultaneously pursue structural transformation, but not all can take advantage of shifting manufacturing activity due to lack of comparative advantage (example: Brazil’s experience).
- Least favorable scenario:
  - A country lacks comparative advantage and becomes a net importer of manufactures.
  - Such a country may finance imports by exporting agricultural goods (despite low productivity) or natural resources, thereby failing to exploit industrialization opportunities.
  - Examples cited: Ghana, Malawi, and Botswana in sub-Sahara Africa (SSA).

*Source: wp18223 - 3.1 Opportunities and Challenges on the Demand Side*

### 4.2 Premature Deindustrialization: A Summary

### 4.2 Premature Deindustrialization: A Summary

### Global patterns and empirical findings
- Felipe and Mehta (2016) — based on data from 64 countries representing 82 percent of the world’s population — find that:
  - The global share of total employment concentrated in manufacturing remained relatively stable at approximately 14 percent.
  - Manufacturing’s share of output remained nearly unchanged throughout this period at approximately 17 percent.
  - Average manufacturing productivity globally has tended to mirror global productivity across all sectors.
- Global stability of manufacturing employment and output shares from 1970 to 2015 implies intense competition for a stable share of world manufactured output; not all countries can increase their individual share.
- Rodrik (2015) on “premature deindustrialization”:
  - Countries that began deindustrializing since 1990 had a peak of manufacturing’s share of employment of 18.9 percent at a per capita income level of 4,273 constant 1990 dollars.
  - Countries that began deindustrializing prior to 1990 peaked at an employment share of 21.5 percent and a per capita income level of 11,048 constant 1990 dollars.
  - Premature deindustrialization implies lower manufacturing employment shares and earlier peaks in per capita income, retarding productivity growth as labor is absorbed by lower-productivity agriculture and services.

### Country case studies: Ghana, Nigeria, Botswana
- Ghana:
  - Maintained an annual growth rate of about 5% over 2001-2010.
  - Structural break beginning in 1992 with intensified movement of labor from agriculture to services and a sharp reduction in manufacturing activity.
  - Manufacturing share fell to a low of 5.14% in 2014 with a slight recovery to 5.33% in 2015.
  - Most labor moving into services was employed in low- and average-productivity services (productivity of one-tenth and one-half of that of high-productivity services).
- Nigeria:
  - Grew at a rate of 6.8% over the last decade (Ajakaiye et al., 2016).
  - Oil sector accounts for 20%-30% of GDP, 70% of government revenues, and 85% of exports.
  - Services share rose from 18% in 1970 to 44% in 2014.
  - Manufacturing share fell from 12% to 6% over the same period.
  - Economy characterized as “transforming from an agrarian economy to a tertiary service economy without undergoing the intermediate stage of industrialization.”
- Botswana:
  - GDP grew at 12 percent per year from 1960 to 1990 and joined upper-middle-income status in 2005.
  - Manufacturing share of GDP rose slowly from 1.4% in 1960 to 6.6% in 2010.
  - Services share rose from 8.5% to 50.6% of GDP over the same period.
  - Share of wholesale and retail trade in GDP rose by 18.2 percentage points.
  - Performance attributed to geographic disadvantage (landlocked), close integration with South Africa, and trade liberalization via SACU.

### Causes of premature deindustrialization
- Global constraint:
  - Stable global shares of manufacturing employment and output reduce the ability of many countries to rely on expanding international demand for manufactured goods.
- Public fundamentals and complementarity:
  - Foreign capital, technology, and skills can substitute for domestic scarcity but rely on complementary public inputs — broadly infrastructure and institutions — to strengthen comparative advantage in manufacturing.
  - Poor infrastructure (e.g., many SSA countries) has deterred investment and upgrading of manufacturing facilities.
- Globalization and supply-chain reallocations:
  - Open global markets allow capital and technology to move to countries with stronger public fundamentals, magnifying manufacturing advantages for those countries.
  - Reallocation of supply chains has benefited countries that can integrate as producers of components without building vertically integrated domestic industries.
- Consequence for excluded countries:
  - Countries lacking strong public fundamentals (or private fundamentals to compensate) become importers of manufactures, fail to exploit industrialization benefits, and terminate structural transformation prematurely.
  - Labor remains in agriculture or moves to lower-productivity services, depriving countries of the productivity gains associated with industrial employment.

### Consequences for productivity and growth
- Industrial sector in low-income industrializing countries can sustain prolonged productivity growth via unconditional cross-country convergence in productivity within industrial sectors.
  - Rodrik (2012) estimates an average rate of convergence in manufacturing industries of 2.9 percent per year, which would translate into an additional boost to productivity of over 6 percent for the industries in the bottom 20 percentile of his sample.
- Premature deindustrialization prevents channeling sufficient employment into manufacturing “escalator” industries, slowing convergence and leaving income per capita on a lower growth trajectory; in some cases this has approached stagnation.
- The rise of labor-saving technology (e.g., AI-enhanced robotics) may further compound difficulties for late industrializers.

### Middle-income trap, malaise, and links to deindustrialization
- Many middle-income countries (e.g., Brazil, Malaysia, Mexico, Peru) have failed to move steadily up the value chain to high-income status.
- Middle-income trap view:
  - Characterized as a “bad, but stable equilibrium” requiring external shock or significant policy intervention to escape.
  - These countries face competition from low-wage low-income countries in mature industries and from high-income countries in high-skill innovation.
- Alternative interpretation:
  - Han and Wei (2016) find LICs take longer to attain middle-income status than middle-income economies take to attain high-income status and find heterogeneity across countries; not all evidence supports an inevitable trap.
- Empirical evidence of TFP slowdown:
  - Many middle-income economies experienced a slowdown in total factor productivity (TFP) that retarded growth, illustrated for Brazil, Argentina, Peru, and Mexico in the 1980-1990 decade.
  - Import-substitution strategies in Brazil and Mexico contributed to a slowdown by limiting quality and competitiveness of manufactured goods and exhausting domestic demand.

### Policy-relevant implications and paths forward
- Essential conditions for successful industrialization and structural transformation:
  - Strong public fundamentals: infrastructure and institutions to attract and leverage foreign capital, technology, and skills.
  - Openness and global competitiveness to ensure demand for tradable manufactured goods is not restricted by low domestic incomes.
  - Complementarities and public policies to stimulate buildup of requisite physical capital and infrastructure for manufacturing expansion.
- Possible medium- to long-term opportunity:
  - Global manufacturing and employment shares have been stable for decades; if large economies (e.g., China) transition successfully to post-industrial modern services without a middle-income trap, space could open for smaller countries to fill manufacturing roles over the medium to longer term.
  - Success depends on other countries (e.g., Brazil) escaping middle-income malaise and on recipient countries having sufficient public fundamentals to attract production.
- Policy takeaway:
  - Latecomer low- and middle-income countries need to prioritize improvement of public fundamentals, infrastructure, institutional quality, openness, and policies that promote complementarities to capture manufacturing opportunities and avoid premature termination of structural transformation.

*Source: IMF Working Paper — 4.2 Premature Deindustrialization: A Summary*

### 5.2  Can Services Do the Trick?

### 5.2  Can Services Do the Trick?

### Patterns of services-led structural change in latecomers (Africa)
- Countries exhibiting movement of labor into traditional (nontradable) services: Ethiopia, Ghana, Keyna, Malawi, Senegal, Tanzania, and Zambia.
- Wholesale and retail trade sector has seen the largest increase in employment share over the last two decades (1990-2000 and 2000-2010) in almost all of these countries (Enache, Ghani, and O’Connell (2016)).
- Wholesale and retail trade sector has low formalization: formalization rate reported as 5.90% (Osei and Jedwab (2017)).
- Empirical finding: movement of labor into wholesale and retail trade is not correlated across countries with the productivity of this sector — observed both where sector productivity is low (Ghana, Kenya, and Senegal) and where it is high (Ethiopia, Malawi, Tanzania, and Zambia) (Enache et al. (2016)) — ruling out supply-side, productivity-based factors as a major explanatory force.
- International specialization explanation:
  - Income increases are driven by agriculture and resource-extraction industries.
  - Proceeds from exports of primary commodities are used to buy tradable manufactures.
  - This generates increased demand for traditional, nontradable services complementary to manufactures.
- Distribution cost evidence: distribution costs are very large for the average consumer good — more than 40% of the retail price in the US and roughly 60% of the retail price in Argentina (Burstein, Neves, and Rebelo (2003)).
- Assessment: movement into traditional nontradable services may generate growth (e.g., unconditional cross-country convergence in services documented in World Economic Outlook, April, 2018), but is unlikely to be a source of long-term growth because of its inability to generate sustained increases in productivity. A credible services-based alternative must rely on growth in high-productivity, tradable services.

### Role of high-productivity (tradable) services: The Indian experience (5.2.1)
- Historical pattern:
  - 1950s–1960s: rapid industrialization emphasizing import substitution (ISI) focused on domestic consumption; manufacturing concentrated on lower quality goods.
  - Structural transformation in India relied on movement of labor from agriculture to services rather than agriculture to manufacturing.
- Sectoral shares (1960s to early 2000s):
  - Manufacturing employment share: increased from 9.8% to 12.4%.
  - Manufacturing’s share in value added GDP: marginal increase (Ahsan and Mitra, 2017).
  - Wholesale and retail trade employment share: increased from 4.88% to 8.27%.
  - Transport and storage employment share: increased from 2.06% to 3.32%.
  - High-productivity services (finance, insurance, real estate, and business services) share: increased from 0.32% to 1.68% (a five-fold increase).
  - High-productivity services share in 1980-1989: 0.40% (increase mainly since 1990).
- Macroeconomic outcomes:
  - India mitigated lack of continued industrialization and maintained a secular increase in per capita GDP over 40 years since it first attained the level of GDP per capita of $800 U.S. (in 2000 dollars).
  - Per capita GDP growth: from 1990 to 2015 grew at 4.81%; further acceleration to 5.63% since 2000.
- Drivers of India’s services-led growth:
  - Longstanding emphasis on scientific and technical knowledge and manpower since independence in 1947.
  - Large pool of technically proficient workers fluent in English (historical accident of British rule).
  - Large Indian diaspora in world centers of information technology, providing networks and tailwinds (e.g., during Y2K).
  - Liberalization of information and communications sectors by government beginning in the mid-1980s.
  - IT sector development was not constrained by the License-Permit Raj affecting traditional industry.
  - IT revolution could bypass inferior physical infrastructure that hindered industry.
- Limits in labor absorption and structural transformation:
  - Employment in high-productivity services increased by 1.28 percentage points.
  - Employment in other services (construction, wholesale and retail, transport and storage; excluding community and social services) increased by 4.91 percentage points.
  - Productivity in other services is one-third to one-fifth of that of high-productivity (communication and IT) counterpart.
  - Conclusion: despite substantial growth and productivity in IT/services, the sector’s ability to absorb large amounts of labor released from agriculture (and manufacturing) was limited in India.

### High-productivity services: potential and prerequisites (5.2.2)
- Private fundamentals required:
  - Access to appropriate technological knowhow and availability of technical manpower are necessary preconditions.
  - India’s case shows strong private fundamentals are critical; without them the process would not commence.
- Public fundamentals and policy roles:
  - Minimum threshold of communication and IT infrastructure is needed to enable the private sector to capitalize on private fundamentals.
  - Government policies that connect domestic activity to international markets (for example, through encouragement of FDI) are desirable in most countries.
  - India’s favorable diaspora networks reduced the need for some government measures.
- Trade-offs between services-led and industrialization-led strategies:
  - Public fundamentals: requirements are less onerous for growth powered by high-productivity services than for industrialization.
  - Private fundamentals: services-led growth requires stronger private fundamentals (technology and skills) that are farther removed from unskilled labor than those needed for manufacturing.
- Fundamental constraint:
  - Services-led growth often requires leapfrogging the production structure relative to domestic demand; it places the burden of production and employment increases on global demand and exports, potentially limiting upside in most cases.
- Overall assessment: optimism about services-led growth is warranted in some contexts, but its potential to engage a large share of labor released by shrinking agriculture (and industry) remains unproven.

### Need for a longer-term perspective in policy making (5.3)
- Traditional focus for low-income countries: accumulation of capital and improvement of infrastructure to strengthen private fundamentals (technology, skills, innovation).
- Latecomers to development, deprived of manufacturing-driven virtuous cycles, risk being denied opportunities that manufacturing historically provided.
- Services-led growth prognosis is not uniformly sanguine; deliberate accumulation of private fundamentals with strong public policy support is required.
- Longer-term policy perspective is necessary because private fundamentals evolve slowly.
- Problems faced by developing countries in the middle-income trap (based on Agenor (2016)) include:
  1. inadequate infrastructure,
  2. diminishing marginal benefits from physical capital investments,
  3. exhaustion of cheap labor and imitation gains,
  4. insufficient human capital to compete with high income countries in innovation,
  5. misallocation of talent, with too much human capital devoted to the production of goods in low growth sectors of the economy,
  6. lack of access to funding needed to finance risky investments in innovative technologies, and
  7. income inequality inhibiting the acquisition of human capital through limited access of low income households to educational opportunities.
- Policy implications:
  - Over a longer horizon, lack of appropriate skills and innovation are key challenges.
  - Public policy should be broad-based, going beyond infrastructure, FDI attraction, and business climate improvements.
  - Policies should address market failures specific to each country and incentivize:
    - individuals to invest in skill upgradation, and
    - businesses to invest in workforce skill upgradation and innovation,
  - These measures aim to sustain long-run growth and prevent stagnation at middle-income levels.

### Conclusion (section 6)
- Manufacturing historically drove productivity growth and large-scale employment absorption, but it may no longer deliver similar outcomes for many low-income countries.
- Many sub-Saharan African countries experienced growth accelerations in the 1990s and 2000s without corresponding manufacturing growth.
- Alternatives:
  - Improving agricultural productivity, or
  - Transitioning to services (including demand for non-tradable services in urbanizing “mega (consumption) cities”).
- Services option:
  - May be viable for some countries, especially with idiosyncratic favorable factors, but unlikely to be a universal model.
  - A few countries have grown rapidly by concentrating on high-productivity tradable services, but the potential for broad structural transformation and large-scale labor absorption is unclear.
- Policy emphasis:
  - Manufacturing can absorb large amounts of labor and catalyze workforce skill upgrades, technology improvement, and product/process innovation.
  - In the absence of strong private fundamentals, public policy must create incentives to strengthen technology, skills, and innovation.
  - Because private fundamentals evolve slowly, public policy should emphasize long-run, sustainable economic growth and structural transformation early in the development process.

*Source: IMF Working Paper excerpt — "5.2  Can Services Do the Trick?"*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18223.pdf_