## wpiea2019263-print-pdf

## Source details

**Canonical URL:** [wpiea2019263-print-pdf](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019263-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019263-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019263-print-pdf.pdf.json)

---

### Convergence success and laggers
- South Korea: GDP per capita rose from 11 percent of the U.S. level to about 70 percent in 2016.
- South Korea headline growth slowed from 10 percent in the 1980s to 7 percent in the 1990s and to 3 percent recently.
- Latin America examples: Brazil, Mexico and Peru grew to middle-income levels and then stagnated for several decades due to significant slowdown in productivity growth.
- Rodrik (2016): Productivity slowdown in some Latin American countries was mainly driven by premature deindustrialization, which dampened aggregate productivity despite within-sector productivity improvements.

### Why China is unlikely to follow the East Asian tiger path
- Benchmark: At China’s current income level, South Korea and Taiwan POC were still growing at 8 percent.
- Two primary reasons China is unlikely to replicate such high growth:
  - Changing global environment:
    - Pre-crisis global growth average: 5.1 percent.
    - Post-crisis global growth average: 3.4 percent.
    - Factors: stagnation in total factor productivity (TFP) growth, rapid aging, slow progress in further global trade integration.
    - Summers (2015): industrial world could be in secular stagnation for a considerable time.
  - Size of the economy:
    - South Korea economy today: about 10 percent of U.S. economy.
    - Taiwan POC economy today: about 6 percent of U.S. economy.
    - Japan economy today: about a quarter of the size of the U.S. economy.
    - China at per capita income of 18 percent of U.S. level: size of the Chinese economy is already 76 percent of the U.S. economy.
    - Implication: World may find it increasingly difficult to absorb ever-increasing exports from China; more demand will need to be generated domestically.

### Why China is unlikely to converge to the global average
- Barro (2016): At China’s current income level, most developing economies were growing at 3-4 percent; mean reversion would suggest China grows at the same rate.
- Offsetting factors for China:
  - Continued political stability, reform and opening up, high savings, strong government investment in infrastructure—these may support growth above the global average.
  - Structural shift from industry to services in China broadly in line with past advanced-economy experiences, mitigating risks from “premature deindustrialization”.

### Approach: supply-side with sectoral transition
- Rationale:
  - Supply-side approach with sectoral transition gives better insight into medium-term growth for a fast-transition economy like China.
  - Sectoral analysis is critical because transition between sectors is a major driver of overall growth.
- Two-stage stylized structural transformation:
  - Stage 1: agriculture share falls; industry and services rise; labor relocates from low-productivity agriculture to higher-productivity manufacturing and services—growth accelerates.
  - Stage 2: at higher income, industry share peaks and declines; services share continues to rise; employment shifts from higher-productivity industry to lower-productivity services—downward pressure on aggregate growth.
- Within-sector upgrades (moving from low-productivity to high-productivity subsectors) are key buffers to moderate slowdown.
- Reform implementation (e.g., SOE reform) can provide sizable boosts to productivity growth (evidence cited from Lam et al. (2017), Zhang (2016), Lardy (2016)).

### Macro overview: sectoral shift and productivity trends
- Two stages of China’s sectoral shift:
  - First stage (Industrialization, 1965–2012):
    - Share of agriculture in GDP fell from 50 percent in the early 1950s to less than 10 percent.
    - Agricultural labor productivity surged tenfold.
    - Overall labor productivity growth was above 9 percent in most years.
  - Second stage (Deindustrialization, 2012–present):
    - Industry's share in GDP peaked at 46 percent in 2011 and has since declined.
    - Industrial labor productivity is 30 percent higher than service productivity and is growing faster.
    - Shift from industry to services exerts a drag on total labor productivity; downward pressure will increase as the productivity gap between industry and services grows.
- Timing: China’s industrial share peaked in 2012 at GDP per capita level of US$10,000 (in 1990 international prices); this timing aligns with past experiences of high-income countries.

### Cross-country deindustrialization benchmarks (selected)
- United States: peak year 1953; GDP/Capita (in 1990 international prices) 10,613; industry share of output at peak 33; service share of output at peak 60.
- UK: peak year 1960; GDP/Capita 8,645; industry share 49; service share 44.
- Japan: peak year 1970; GDP/Capita 9,714; industry share 42; service share 51.
- Taiwan POC: peak year 1986; GDP/Capita 7,477; industry share 49; service share 44.
- Korea: peak year 1992; GDP/Capita 9,877; industry share 44; service share 47.
- China: peak year 2012; GDP/Capita 9,719; industry share 46; service share 46.

### Services and industry shares and dynamics
- Nominal share of services in GDP in China: 52 percent.
- Comparators: services share roughly 70-80 percent in the U.S., Japan and Australia; 60 percent in South Korea.
- Implication: significant room for services catch-up, especially in interior regions (Beijing and Shanghai already at advanced-economy levels).
- Industrial productivity:
  - Manufacturing labor productivity growth around 8 percent, supported by capital deepening and industrial upgrading.
  - TFP growth in industry slowed in the past decade but stabilized in recent years.
- Services productivity:
  - Services labor productivity growth slowed from about 6 percent to less than 3 percent since deindustrialization started in 2012.
  - TFP growth in services has shown a declining trend since early 2000.
  - Contributing factor: more than eight million workers laid off from industry were largely absorbed into low-skill service jobs, lowering overall services productivity growth.

### Productivity levels and convergence potential
- Industrial labor productivity: rose from about 13 percent of the U.S. level in 1997 to 33 percent in 2015.
- Services labor productivity: rose from nearly 10 percent to 29 percent over the same period.
- Conclusion: despite significant catch-up, substantial room remains for further convergence in both industry and services.

### Demographics
- Working-age population:
  - Peak in 2010.
  - Expected fall from 1,004 million in 2015 to 970 million by 2030, with pace of decline accelerating over time.
- Near-term impact: limited, as labor force remains stable due to rising labor participation rate and increased average working hours.
- Medium-term impact: shrinking working population will lead to a falling labor force, exerting downward pressure on growth.
- Human capital: has risen steadily but the pace of increase has slowed and is expected to slow further; not likely to offset population aging effects.

### Industrial analysis: sectoral shift from low-tech to high-tech
- Classification (38 two-digit industrial sectors into OECD R&D intensity categories):
  - Low-tech: consumer goods (apparel, food and beverages, etc.).
  - Medium-tech: metal and chemical products.
  - High-tech: ICT, machinery, medicine, etc.
- Real industrial value added shares (1990 domestic prices):
  - High-tech share rose from 24 percent in 1980 to 52 percent in 2015.
  - Medium-tech share fell from 48 percent to 26 percent.
  - Low-tech share stable around 30 percent from 1980 to 1994, then fell to 20 percent in 2015.
- Employment shares:
  - High-tech employment share rose from 28 percent in the 1990s to 43 percent today.
  - Medium-tech employment share fell from close to 40 percent in 2000 to slightly above 30 percent.
  - Low-tech employment share rose in the early 2000s then declined as upgrading dominated.
- Nominal value added: sectoral shift less pronounced in nominal terms due to relative price changes (ICT prices declined significantly since 2000; prices of low- and medium-tech products rose).

### China’s advanced industrial structure
- In 2015 (2011 international dollar prices), share of high-tech in industrial real value added was 43 percent, similar to levels in Belgium and Spain despite China’s lower income.

### Productivity within subsectors
- Labor productivity growth (average annual rates, 1996–2015):
  - High-tech industries: 16 percent.
  - Medium-tech industries: 14 percent.
  - Low-tech industries: 13 percent.
- Implication: high-tech labor productivity is twice that of medium-tech and nearly three times that of low-tech industries.
- TFP divergence:
  - Since mid-1990s, TFP growth in high-tech industries increased fivefold.
  - TFP growth less than doubled in low-tech and medium-tech sectors.
  - Medium-tech sectors have lower TFP levels than low-tech sectors despite higher labor productivity—likely reflecting larger contribution from capital stock.
- Post-global financial crisis: temporary slowdown across industrial subsectors, followed by rebound—particularly strong in high-tech.
- Convergence status by sector (labor productivity as share of frontier, 2000 to 2015):
  - Low-tech industry: rose from 13 percent to above 55 percent.
  - Medium-tech industry: rose from 10 percent to 38 percent.
  - High-tech industry: rose from 11 percent to 40 percent.
- Conclusion: Productivity convergence is more advanced in low-tech sectors; ample room remains for catch-up across all industries.

### Box 2 — ICT sector: driver of high-tech
- ICT role and recent performance:
  - ICT is a key component of high-tech.
  - ICT industry experienced the fastest productivity growth of all high-tech sectors over the past two decades.
  - ICT share in high-tech real value added rose from 30 percent at the end 1990s to about 42 percent in 2015.
  - In 2015, labor productivity in the ICT sector was twice as high as in other high-tech sectors.
  - Continued resource allocation to ICT provided a significant boost to overall high-tech productivity growth, although such allocation has slowed recently.
- Convergence dynamics and specialization:
  - ICT convergence lagged other high-tech sectors since 2000; ICT labor productivity as a share of the frontier fell from 2000 and picked up only since 2013.
  - Slower ICT convergence since 2000 likely reflects China’s specialization in less sophisticated production stages (assembly) following WTO entry; recent years show resumed ICT convergence due to movement into more advanced ICT production and surging R&D.
- Services context relevant to ICT:
  - Overall services productivity converged from 10 percent of the frontier in the mid-1990s to 29 percent in 2015, lagging industrial convergence.
  - Services classification (OECD) divides 11 subsectors into non-market and market services (listed in source).
- Historical phases of sectoral shift (1980–present):
  - First phase (1980–1990): market services nominal share rose from 70 percent to 80 percent; employment share in market services increased.
  - Second phase (1990–2002): nominal share of market services retreated from 80 percent to 72 percent; non-market services expanded.
  - Third phase (2003–present): since WTO entry in 2001, market services (business services, wholesale and financial services) expanded; employment share of market services broadly stable at 70 percent until 2012, then rose.
- Expected structural evolution and implications for ICT:
  - Cross-country experience: market services share may decline as income increases; China’s market services share peaked in 2012 at 75 percent and declined thereafter.
  - Market services share in China expected to decline from 74 percent currently to 65-70 percent.
  - Wholesale and retail sales account for 35 percent of total services employment in China versus 15 percent in the U.S.; health employment 3 percent in China versus 17 percent in the U.S.
  - Improving productivity in wholesale and retail could free labor for healthcare and education.
- Productivity within services and ICT-related subsectors:
  - From 1980 to end-1990s, market and non-market services productivity grew at similar rates of 5-6 percent.
  - Since early 2000, market services productivity accelerated to 7.3 percent, non-market services slowed to 4.8 percent.
  - Since 2012, productivity growth of market services slowed, likely due to absorption of low-skilled laid-off industrial workers.
  - Leaders since early 2000s: financial services (productivity surged threefold by 2015); transportation and ICT services (productivity doubled).
  - Lagging market subsectors: wholesale and retail sales; business services.
  - Non-market services: education and healthcare productivity growth accelerated to 14 percent since 2003; cultural, sport and entertainment, household services stable at 4 percent; public administration productivity has fallen in recent years.
- Convergence levels (market services):
  - Overall market services productivity in China about 21 percent of the U.S. frontier.
  - Financial services reportedly higher than most advanced economies (measurement issues noted).
  - Transportation convergence at 40 percent of frontier.
  - ICT sector convergence close to 30 percent of frontier.
  - Business services and hotels/restaurants around 25 percent of frontier.
  - Wholesale and real estate services below 15 percent of frontier (likely reflecting restrictive regulation).
- Convergence of non-market services:
  - Labor productivity of non-market services reached 35 percent of the U.S. level and 48 percent of the level in other developed countries (such as Germany).
  - Convergence mostly driven by healthcare and education; measurement uncertainty noted.
- Implication: China can maintain robust growth in the coming decade but on a gradual downward trend due to structural transition to services and shrinking distance to the frontier.

### Growth projections and baseline (sectoral convergence forecasts)
- GDP growth projections:
  - GDP growth will slow from 6.5 percent currently to around 5 percent in 2025 and to 4-4.2 percent by 2030.
- Productivity convergence:
  - China’s productivity is forecast to converge to 45 percent of the U.S. level by 2030 (from 30 percent in 2015).
- Sectoral convergence (industry and services):
  - Industrial sector: convergence to 56 percent of frontier from 35 percent currently.
    - Within industry:
      - Low-tech sectors forecast to converge to around 80 percent of frontier.
      - Mid-tech sectors projected to reach about 50 percent of frontier.
      - High-tech sectors (ICT) projected to reach 40 percent of frontier.
  - Services sector: convergence expected to reach 44 percent of frontier, from 25 percent currently.
    - Within services:
      - Education, health, and transportation would reach 65 percent of frontier.
      - Business services, real estate services, and government administration may converge slowly and remain below 30 percent of the U.S. level.

### Baseline aggregate and sectoral forecasts to 2030
- Sectoral shares by 2030 (baseline bottom-up subsector convergence models):
  - Services: 65 percent in value added and 51 percent in employment.
  - Industry: 30 percent in value added and 25 percent in employment.
  - Agriculture: 5 percent in value added and 23 percent in employment.
  - Within industry: high-tech sectors share estimated to rise to 50 percent by 2030; low-tech sectors fall to 30 percent.
  - For services: non-market services likely to rise to nearly 30 percent by 2030, driven by education and healthcare.
  - Within market services: ICT, finance, and business services expected to gain importance; shares of wholesale and retail, transportation, and hotels expected to decline.
- Aggregate labor productivity growth forecast: slow from 6.6 percent in 2018 to 4.2 percent in 2030.
- Structural shift effects:
  - Historically, structural shifts ("between effect") contributed about 25 percent to overall productivity growth in the past three decades.
  - Going forward, relocation from industry to services is expected to put downward pressure on productivity growth, with the contribution of the between effect diminishing to zero.
  - Future productivity growth will depend primarily on within-sector upgrading.

### Reform scenarios and estimated impacts
- SOE reform:
  - Current facts:
    - SOEs account for 38 percent of total industrial assets.
    - In 2017, ROA of industrial SOEs was 4 percent, compared with 10 percent for private firms.
    - Asset size of SOEs stands at 60 percent of GDP.
  - Potential gains:
    - Raising ROA of industrial SOEs to 6 percent could boost GDP growth by around 1 percentage point.
    - Firm-level evidence (Lam et al. (2017)): SOE reforms could raise productivity and GDP growth by 0.7-1.2 percent per year; policies include proactive exits of zombie firms and hardening budget constraints of SOEs.
    - With faster SOE reform, industrial convergence could reach 66 percent of frontier by 2030 (versus 56 percent baseline).
    - Faster SOE reform could boost annual potential growth by about 1 percentage point.
- Opening up services:
  - Evidence: services productivity convergence highly correlated with opening to foreign entry.
  - China’s services sector remains very restrictive; business services productivity stands at only 17 percent of the frontier level.
  - Scenario:
    - If services sector restrictions in China are lifted to the OECD average by 2030, overall services productivity would improve to 57 percent of frontier, compared with 47 percent in the baseline.
    - This liberalization would boost annual GDP growth by around 0.5 percentage points.
    - Summary notes lifting regulations to the OECD average could bring convergence to 57 percent of frontier by 2030 and boost annual growth by 0.4-0.7 percentage points.
  - Aggregate contribution: greater opening in services could increase growth by around 0.5 percentage points.
- Combined reforms:
  - If both faster SOE reform and greater opening in services are implemented, potential GDP growth could reach up to 5½ percent.

### Digitalization: scope, effects, and policy
- Digitalization metrics:
  - Measured broadly, about 33 percent of China’s GDP is digitalized.
  - The size of the digital economy has been expanding at above 15 percent annually.
- Empirical association:
  - Chen and Zhang (2018) estimate a 1-percentage-point increase in overall digitalization is associated with an increase of 0.3 percentage points in GDP growth, with a two-year lag.
  - A slower pace of digitalization could therefore slow growth by 1-4 percentage points.
- Drivers and policy:
  - Success driven by private sector dynamism; key to continued success is promoting healthy development of private companies.
  - Government support recommended: invest in digital infrastructure, implement active labor and education policy to prepare for technological change, and strengthen the social safety net to protect laid-off workers during the transition.
- Baseline assumption: continued high-speed digitalization underlies the baseline growth path.

### Global trade integration, adverse scenarios, and risks
- Reversal of global trade and technology integration would hinder productivity convergence via reduced technology transfer and assimilation.
- Two illustrative adverse scenarios for convergence speed:
  - Convergence speed at 80 percent of the baseline → potential growth in China would slow to below 4 percent by 2030.
  - Convergence speed at 50 percent of the baseline → potential growth in China would slow to below 3 percent by 2030.
- Policy implication: resolve trade tensions under a multilateral framework and reform the global trading system.

### Financial vulnerabilities and other risks
- After years of rapid credit expansion, financial vulnerabilities have risen and the risk of a financial crisis has increased.
- Since 2017, deleveraging and regulatory tightening have slowed credit growth and reduced interconnectedness between banks and non-bank financial institutions.
- Nonetheless, vulnerabilities remain elevated and further efforts are needed to unwind accumulated financial risks to avoid an abrupt adjustment with long-lasting growth impacts.

### Key summary conclusions and policy priorities
- Historical context: per capita income rose from 5 percent of the U.S. level to 30 percent; 800 million people were lifted out of poverty.
- Baseline outlook: China’s potential growth will slow but remain robust at around 4 percent by 2030 under the baseline.
- Reform potential:
  - Faster SOE reform could boost annual potential growth by about 1 percentage point.
  - Greater opening in services could increase growth by around 0.5 percentage points.
  - If both reforms are implemented, potential GDP growth could reach up to 5½ percent.
- Headwinds: potential reversal in global trade and technology integration and the risk of an abrupt financial adjustment could lower 2030 growth to about 3 percent.
- Policy priorities:
  - Advance SOE reforms (zombie firm exits, harder budget constraints).
  - Liberalize services to OECD-average entry conditions and attract foreign expertise.
  - Promote digitalization via private sector dynamism, digital infrastructure investment, active labor and education policies, and a stronger social safety net.
  - Resolve China-U.S. tensions under a multilateral framework and continue domestic deleveraging reforms.

*Source: wpiea2019263-print-pdf*

### Box 1: Convergence success and laggers

### Box 1: Convergence success and laggers

### Convergence success and laggers
- South Korea: GDP per capita rose from 11 percent of the U.S. level to about 70 percent in 2016.
- South Korea headline growth slowed from 10 percent in the 1980s to 7 percent in the 1990s and to 3 percent recently.
- Latin America (examples): Brazil, Mexico and Peru grew to middle-income levels and then stagnated for several decades due to significant slowdown in productivity growth.
- Rodrik (2016): Productivity slowdown in some Latin American countries was mainly driven by premature deindustrialization, which dampened aggregate productivity despite within-sector productivity improvements.

### Why China is unlikely to follow the East Asian tiger path
- Benchmark comparison: When at China’s current income level, South Korea and Taiwan POC were still growing at 8 percent.
- Two primary reasons China is unlikely to replicate such high growth:
  - Changing global environment:
    - Pre-crisis global growth average: 5.1 percent.
    - Post-crisis global growth average: 3.4 percent.
    - Factors: stagnation in total factor productivity (TFP) growth, rapid aging, slow progress in further global trade integration.
    - Summers (2015): industrial world could be in secular stagnation for a considerable time.
  - Size of the economy:
    - South Korea economy today: about 10 percent of U.S. economy.
    - Taiwan POC economy today: about 6 percent of U.S. economy.
    - Japan economy today: about a quarter of the size of the U.S. economy.
    - China at per capita income of 18 percent of U.S. level: size of the Chinese economy is already 76 percent of the U.S. economy.
    - Implication: World may find it increasingly difficult to absorb ever-increasing exports from China; more demand will need to be generated domestically.

### Why China is unlikely to converge to the global average
- Barro (2016): At China’s current income level, most developing economies were growing at 3-4 percent; mean reversion would suggest China grows at the same rate.
- Offsetting factors for China:
  - Continued political stability, reform and opening up, high savings, strong government investment in infrastructure—these may support growth above the global average.
  - Structural shift from industry to services in China broadly in line with past advanced-economy experiences, mitigating risks from “premature deindustrialization”.

### Approach: supply-side with sectoral transition
- Rationale:
  - Supply-side approach with sectoral transition gives better insight into medium-term growth for a fast-transition economy like China.
  - Sectoral analysis is critical because transition between sectors is a major driver of overall growth.
- Two-stage stylized structural transformation:
  - Stage 1: Share of agriculture falls; industry and services rise; labor relocates from low-productivity agriculture to higher-productivity manufacturing and services—growth accelerates.
  - Stage 2: At higher income, industry share peaks and declines; services share continues to rise; employment shifts from higher-productivity industry to lower-productivity services—downward pressure on aggregate growth.
- Upgrades within industry and services (moving from low-productivity to high-productivity subsectors) are key buffers to moderate slowdown.
- Reform implementation (e.g., SOE reform) can provide sizable boosts to productivity growth (evidence from Lam et al. (2017), Zhang (2016), Lardy (2016)).

### Macro overview: sectoral shift and productivity trends
- Two stages of China’s sectoral shift:
  - First stage (Industrialization, 1965–2012):
    - Share of agriculture in GDP fell from 50 percent in the early 1950s to less than 10 percent.
    - Agricultural labor productivity surged tenfold.
    - Overall labor productivity growth was above 9 percent in most years.
  - Second stage (Deindustrialization, 2012–present):
    - Industry's share in GDP peaked at 46 percent in 2011 and has since declined.
    - Industrial labor productivity is 30 percent higher than service productivity and is growing faster.
    - Shift from industry to services exerts a drag on total labor productivity; downward pressure will increase as the productivity gap between industry and services grows.
- Timing:
  - China’s industrial share peaked in 2012 at GDP per capita level of US$10,000 (in 1990 international prices); this timing aligns with past experiences of high-income countries, reducing concerns of premature deindustrialization.

### Cross-country deindustrialization benchmarks (selected entries from table)
- United States: peak year 1953; GDP/Capita (in 1990 international prices) 10,613; industry share of output at peak 33; service share of output at peak 60.
- UK: peak year 1960; GDP/Capita 8,645; industry share 49; service share 44.
- Japan: peak year 1970; GDP/Capita 9,714; industry share 42; service share 51.
- Taiwan POC: peak year 1986; GDP/Capita 7,477; industry share 49; service share 44.
- Korea: peak year 1992; GDP/Capita 9,877; industry share 44; service share 47.
- China: peak year 2012; GDP/Capita 9,719; industry share 46; service share 46.

### Services and industry shares and dynamics
- Nominal share of services in GDP in China: 52 percent.
- Comparators: services share roughly 70-80 percent in the U.S., Japan and Australia; 60 percent in South Korea.
- Implication: significant room for services catch-up, especially in interior regions (Beijing and Shanghai already at advanced-economy levels).
- Industrial productivity:
  - Manufacturing labor productivity growth around 8 percent, supported by capital deepening and industrial upgrading.
  - TFP growth in industry slowed in the past decade but stabilized in recent years.
- Services productivity:
  - Services labor productivity growth slowed from about 6 percent to less than 3 percent since deindustrialization started in 2012.
  - TFP growth in services has shown a declining trend since early 2000.
  - Contributing factor: more than eight million workers laid off from industry were largely absorbed into low-skill service jobs, lowering overall services productivity growth.

### Productivity levels and convergence potential
- Industrial labor productivity:
  - Rose from about 13 percent of the U.S. level in 1997 to 33 percent in 2015.
- Services labor productivity:
  - Rose from nearly 10 percent to 29 percent over the same period.
- Conclusion: despite significant catch-up, substantial room remains for further convergence in both industry and services.

### Demographics
- Working-age population:
  - Peak in 2010.
  - Expected fall from 1,004 million in 2015 to 970 million by 2030, with pace of decline accelerating over time.
- Near-term impact:
  - Limited, as labor force remains stable due to rising labor participation rate and increased average working hours.
- Medium-term impact:
  - Shrinking working population will lead to a falling labor force, exerting downward pressure on growth.
- Human capital:
  - Has risen steadily but the pace of increase has slowed and is expected to slow further; not likely to offset population aging effects.

### Industrial analysis: sectoral shift from low-tech to high-tech
- Classification of 38 two-digit industrial sectors into three categories (OECD R&D intensity):
  - Low-tech: consumer goods (apparel, food and beverages, etc.).
  - Medium-tech: metal and chemical products.
  - High-tech: ICT, machinery, medicine, etc.
- Real industrial value added shares (1990 domestic prices):
  - High-tech share rose from 24 percent in 1980 to 52 percent in 2015.
  - Medium-tech share fell from 48 percent to 26 percent.
  - Low-tech share stable around 30 percent from 1980 to 1994, then fell to 20 percent in 2015.
- Employment shares:
  - High-tech employment share rose from 28 percent in the 1990s to 43 percent today.
  - Medium-tech employment share fell from close to 40 percent in 2000 to slightly above 30 percent.
  - Low-tech employment share rose in the early 2000s (WTO entry and booming low-tech exports) then declined as upgrading dominated.
- Nominal value added:
  - Sectoral shift less pronounced in nominal terms due to relative price changes (ICT prices declined significantly since 2000; prices of low- and medium-tech products rose).

### China’s advanced industrial structure
- In 2015 (2011 international dollar prices), share of high-tech in industrial real value added was 43 percent, similar to levels in Belgium and Spain despite China’s lower income.

### Productivity within subsectors
- Labor productivity growth (average annual rates, 1996–2015):
  - High-tech industries: 16 percent.
  - Medium-tech industries: 14 percent.
  - Low-tech industries: 13 percent.
- Implication: high-tech labor productivity is twice that of medium-tech and nearly three times that of low-tech industries.
- TFP divergence:
  - Since mid-1990s, TFP growth in high-tech industries increased fivefold.
  - TFP growth less than doubled in low-tech and medium-tech sectors.
  - Medium-tech sectors have lower TFP levels than low-tech sectors despite higher labor productivity—likely reflecting larger contribution from capital stock.
- Post-global financial crisis:
  - Temporary slowdown across industrial subsectors, followed by rebound—particularly strong in high-tech.
- Convergence status by sector (labor productivity as share of frontier, 2000 to 2015):
  - Low-tech industry: rose from 13 percent to above 55 percent.
  - Medium-tech industry: rose from 10 percent to 38 percent.
  - High-tech industry: rose from 11 percent to 40 percent.
- Conclusion: Productivity convergence is more advanced in low-tech sectors; ample room remains for catch-up across all industries.

*Source: IMF staff compilation of Box 1 content from wpiea2019263-print-pdf*

### Box 2: ICT sector – Driver of High-tech

### Box 2: ICT sector – Driver of High-tech

### Role and recent performance of the ICT sector
- The ICT sector is a key component of the high-tech sector.
- Over the past two decades, the ICT industry has experienced the fastest productivity growth of all high-tech sectors.
- The ICT share in high-tech industries' real value added rose from 30 percent at the end 1990s to about 42 percent in 2015.
- In 2015, labor productivity in the ICT sector was twice as high as in other high-tech sectors.
- Continued resource allocation to the ICT sector provided a significant boost to overall high-tech productivity growth, although such allocation has slowed in recent years.

### Convergence dynamics and sectoral specialization
- The convergence of the ICT sector has lagged other high-tech sectors since 2000:
  - From 2000, labor productivity in the ICT industry as a share of the frontier fell, and has been picking up only since 2013.
  - Other high-tech sectors, such as machinery and medicine, saw productivity rise from around 10 percent of the frontier level to 30-60 percent.
- The slower ICT convergence since 2000 likely reflects China’s specialization in less sophisticated production stages of the global supply chain (e.g., the assembly of cellphones and computers) following WTO entry, while advanced economies focused on chip design and production.
- In more recent years, ICT convergence in China has resumed as the economy moved toward more advanced parts of ICT production, benefiting from surging R&D.

### Services sector context relevant to ICT
- Overall services productivity converged from 10 percent of the frontier in the mid-1990s to 29 percent in 2015, lagging industrial sector convergence.
- Services convergence has been driven by both a sectoral shift from non-market services to market services and productivity increases in subsectors.
- Classification used (OECD definition) divides 11 service subsectors into:
  - Non-market services: ADM (government, public administration, political and social organizations, etc.), HEA (healthcare and social security services), EDU (education), SER (cultural, sports, entertainment services; residential and other services).
  - Market services: HOT (hotels and restaurants), FIN (financial intermediation), P&T (information and computer services), REA (real estate services), SAL (wholesale and retail trades), T&S (transport, storage and post services), BUS (leasing, technical, science and business services).

### Historical phases of sectoral shift (1980–present)
- First phase (1980–1990):
  - Market services nominal share in total services rose from 70 percent to 80 percent.
  - Employment share in market services increased, driven by wholesale and retail sales and the financial sector.
- Second phase (1990–2002):
  - Nominal share of market services retreated from 80 percent to 72 percent.
  - Non-market services expanded, driven by culture, healthcare and education.
  - Employment share of market services increased from 1990–1996 and then declined steadily.
- Third phase (2003–present):
  - Since WTO entry in 2001, market services—particularly business services, wholesale and financial services—expanded, with nominal value added share increasing steadily.
  - Employment share of market services stayed broadly stable at 70 percent until 2012, then rose since then with increasing labor allocation to sales, ICT service and business sectors.

### Expected structural evolution of services and implications for ICT
- Cross-country experience suggests the share of market services may decline as income increases; consumer demand shifts toward non-market services (education, healthcare, sports).
- China’s market services share peaked in 2012 at 75 percent and declined thereafter.
- The share of market services in China is expected to decline from 74 percent currently to 65-70 percent (as seen in the U.S. and European countries).
- Labor reallocation opportunity:
  - Wholesale and retail sales account for 35 percent of total services employment in China versus 15 percent in the U.S.
  - Employment in health accounts for 3 percent in China versus 17 percent in the U.S.
  - Improving productivity in wholesale and retail sales could free labor for healthcare and education.

### Productivity within services and ICT-related subsectors
- Divergence since early 2000:
  - From 1980 to end-1990s, market and non-market services productivity grew at similar rates of 5-6 percent.
  - Since early 2000, market services productivity accelerated to 7.3 percent, while non-market services slowed to 4.8 percent.
  - Since 2012, productivity growth of market services slowed, likely due to absorption of low-skilled laid-off industrial workers.
- Leaders in market services productivity growth since early 2000s:
  - Financial services: productivity surged threefold by 2015.
  - Transportation and ICT services: productivity doubled.
- Lagging market subsectors:
  - Wholesale and retail sales: productivity barely increased from 1978–2003 and has risen only slowly since then.
  - Business services: slow productivity growth.
- Non-market services productivity:
  - Education and healthcare productivity growth accelerated to 14 percent since 2003.
  - Cultural, sport and entertainment, and household services stable at 4 percent.
  - Public administration productivity has fallen in recent years after decades of steady increases (likely reflecting measurement issues).
- Convergence levels (market services):
  - Overall productivity of market services in China is about 21 percent of the U.S. frontier.
  - Financial services in China have higher labor productivity than in most advanced economies based on traditional metrics (measurement issues possible).
  - Transportation convergence at 40 percent of frontier.
  - ICT sector convergence close to 30 percent of frontier.
  - Business services and hotels/restaurants around 25 percent of frontier.
  - Wholesale and real estate services below 15 percent of frontier (likely reflecting restrictive regulation for entry).
- Convergence of non-market services:
  - Labor productivity of non-market services reached 35 percent of the U.S. level and 48 percent of the level in other developed countries (such as Germany).
  - Convergence mostly driven by healthcare and education; cultural and sports sector and government administration lagging.
  - Measurement of non-market services productivity subject to considerable uncertainty due to price subsidies and quality measurement issues.

### Implications for future growth and ICT’s contribution
- China has potential to maintain robust growth in the coming decade but on a gradual downward trend due to structural transition to services and shrinking distance to the frontier.
- Baseline projections (bottom-up approach using subsector convergence models):
  - Industrial and services productivity growth expected to slow to 3-  4 percent by 2030 (applying convergence speeds from Rodrik (2013) and Bourlès et al. (2010)).
  - Sectoral shares by 2030:
    - Services: 65 percent in value added and 51 percent in employment.
    - Industry: 30 percent in value added and 25 percent in employment.
    - Agriculture: 5 percent in value added and 23 percent in employment.
    - Within industry: high-tech sectors share estimated to rise to 50 percent by 2030; low-tech sectors fall to 30 percent.
    - For services: non-market services likely to rise to nearly 30 percent by 2030, driven by education and healthcare.
    - Within market services: ICT, finance, and business services expected to gain importance; shares of wholesale and retail, transportation, and hotels expected to decline.
  - Aggregate labor productivity growth forecast to slow from 6.6 percent in 2018 to 4.2 percent in 2030.
- Structural shift effects:
  - Historically, structural shifts (the "between effect") contributed about 25 percent to overall productivity growth in the past three decades.
  - Going forward, relocation from industry to services is expected to put downward pressure on productivity growth, with the contribution of the between effect diminishing to zero.
  - Future productivity growth will depend primarily on within-sector upgrading.
- Other constraints and risks noted:
  - Aging population and shrinking labor force will put downward pressure on labor supply, though the impact is expected to be only marginal.

*Source: wpiea2019263-print-pdf - Box 2: ICT sector – Driver of High-tech*

### 0.1 percent to the decline in growth. Overall, this would imply that GDP growth will slow

### wpiea2019263-print-pdf - 0.1 percent to the decline in growth. Overall, this would imply that GDP growth will slow

### Growth projections and baseline
- GDP growth will slow from 6.5 percent currently to around 5 percent in 2025 and to 4-4.2 percent by 2030.
- China’s productivity is forecast to converge to 45 percent of the U.S. level by 2030.
- Assuming frontier productivity continues to grow at a stable rate, overall labor productivity in China is likely to reach 45 percent of frontier by 2030, from 30 percent in 2015.

### Sectoral convergence (industry and services)
- Industrial sector:
  - Convergence to 56 percent of frontier from 35 percent currently.
  - Within industry:
    - Low-tech sectors (wood, food and beverages) forecast to converge to around 80 percent of frontier.
    - Mid-tech sectors (chemicals and optical products) projected to reach about 50 percent of frontier.
    - High-tech sectors (ICT) projected to reach 40 percent of frontier.
- Services sector:
  - Convergence expected to reach 44 percent of frontier, from 25 percent currently.
  - Within services:
    - Education, health, and transportation would reach 65 percent of frontier.
    - Business services, real estate services, and government administration may converge slowly and remain below 30 percent of the U.S. level.

### State-owned enterprise (SOE) reform and impact
- Current facts:
  - In the industrial sector, SOEs account for 38 percent of total industrial assets.
  - In 2017, the return on assets (ROA) of industrial SOEs was 4 percent, compared with 10 percent for private firms.
  - Asset size of SOEs stands at 60 percent of GDP.
- Potential gains:
  - Raising the ROA of industrial SOEs to 6 percent could boost GDP growth by around 1 percentage point.
  - Firm-level evidence (Lam et al. (2017)) suggests SOE reforms could raise productivity and GDP growth by 0.7-1.2 percent per year; required policies include more proactive exits of zombie firms and hardening budget constraints of SOEs.
  - With faster productivity growth from SOE reform, industrial convergence could reach 66 percent of frontier by 2030, compared with 56 percent in the baseline.
- Aggregate scenario:
  - Faster SOE reform could boost annual potential growth by about 1 percentage point.

### Opening up services and productivity gains
- Cross-country evidence indicates services productivity convergence is highly correlated with opening to foreign entry.
- China’s services sector remains very restrictive; business services productivity stands at only 17 percent of the frontier level.
- Scenario:
  - If services sector restrictions in China are lifted to the OECD average by 2030, overall services productivity would improve to 57 percent of frontier, compared with 47 percent in the baseline.
  - This liberalization would boost annual GDP growth by around 0.5 percentage points.
  - Summary later notes lifting regulations to the OECD average could bring convergence to 57 percent of frontier by 2030 and boost annual growth by 0.4-0.7 percentage points.
- Aggregate contribution:
  - Greater opening in the service sector could increase growth by around 0.5 percentage points.

### Digitalization: scope, effects, and policy
- Measured broadly, about 33 percent of China’s GDP is digitalized.
- The size of the digital economy has been expanding at above 15 percent annually.
- Empirical association:
  - Chen and Zhang (2018) estimate a 1-percentage-point increase in overall digitalization is associated with an increase of 0.3 percentage points in GDP growth, with a two-year lag.
  - A slower pace of digitalization could therefore slow growth by 1-4 percentage points.
- Drivers and policy:
  - Success driven by private sector dynamism; key to continued success is promoting healthy development of private companies.
  - Government support recommended: invest in digital infrastructure, implement active labor and education policy to prepare for technological change, and strengthen the social safety net to protect laid-off workers during the transition.
- Baseline assumption:
  - Continued high-speed digitalization underlies the baseline growth path.

### Global trade integration, adverse scenarios, and risks
- Reversal of global trade and technology integration would hinder productivity convergence via reduced technology transfer and assimilation.
- Two illustrative adverse scenarios for convergence speed:
  - Convergence speed at 80 percent of the baseline → potential growth in China would slow to below 4 percent by 2030.
  - Convergence speed at 50 percent of the baseline → potential growth in China would slow to below 3 percent by 2030.
- Policy implication:
  - To avoid such scenarios, resolve trade tensions under a multilateral framework and reform the global trading system.

### Risks of a financial crisis
- After years of rapid credit expansion, financial vulnerabilities have risen and the risk of a financial crisis has increased.
- Since 2017, deleveraging and regulatory tightening have slowed credit growth and reduced interconnectedness between banks and non-bank financial institutions.
- Nonetheless, vulnerabilities remain elevated and further efforts are needed to unwind accumulated financial risks to avoid an abrupt adjustment with long-lasting growth impacts.

### Summary conclusions and overall policy implications
- Historical context:
  - Per capita income rose from 5 percent of the U.S. level to 30 percent; 800 million people were lifted out of poverty.
- Baseline outlook:
  - China’s potential growth will slow but remain robust at around 4 percent by 2030 under the baseline.
  - Baseline assumes further digitalization based on continuing private innovation and avoiding excessive government intervention.
- Reform potential:
  - Faster SOE reform could boost annual potential growth by about 1 percentage point.
  - Greater opening in services could increase growth by around 0.5 percentage points.
  - If both reforms are implemented, potential GDP growth could reach up to 5½ percent.
- Headwinds:
  - Potential reversal in global trade and technology integration and the risk of an abrupt financial adjustment could lower 2030 growth to about 3 percent.
- Policy priorities:
  - Advance SOE reforms (zombie firm exits, harder budget constraints).
  - Liberalize services to OECD-average entry conditions and attract foreign expertise.
  - Promote digitalization via private sector dynamism, digital infrastructure investment, active labor and education policies, and a stronger social safety net.
  - Resolve China-U.S. tensions under a multilateral framework and continue domestic deleveraging reforms.

*Source: wpiea2019263-print-pdf - 0.1 percent to the decline in growth. Overall, this would imply that GDP growth will slow*

### References:

### wpiea2019263-print-pdf - References

### References (selected)
- Atkinson, A. B, 2005, Measurement of UK Government Output and Productivity for the National Accounts, Journal of the Statistical and Social Inquiry Society of Ireland, Vol. XXXIV, 152-160
- Bai Chongen, and Zhang Qiong, 2016, A Research on China’s Economic Growth Potential, New Finance Review (in Chinese)
- Barro Robert J., 2015, Convergence and Modernization, The Economic Journal, 125 (June), 911–942
- Baumol, W., 1967, Macroeconomics Of Unbalanced Growth: The Anatomy Of Urban Crisis, The American Economic Review, 57(3), 415-426
- Baumol, W. 1985, Productivity policy and the service sector, in Robert Inman (ed.), Managing the Service Economy: Prospects and Problems, Cambridge University Press
- Barro Robert J., 2016, Economic Growth and Convergence, Applied to China, China & World Economy, 24(5), 5-19
- Bourlès Renaud, Gette Gilbert, Lopez Jimmy, Mairesse Jacques & Nicoletti Giuseppe, 2013, Do Product Market Regulations In Upstream Sectors Curb Productivity Growth? Panel Data Evidence for OECD Countries, The Review of Economics and Statistics, vol. 95(5), 1750-1768
- Borio Claudio, Piti Disyatat and Mikael Juselius, 2013, Rethinking potential output: Embedding information about the financial cycle, BIS Working Papers No.404
- Buera Francisco J. and Kaboski Joseph P., 2012, The Rise of the Service Economy, American Economic Review, 102 (6): 2540-69
- Byrne David M., Fernald John G. and Reinsdorf Marshall B., 2016, Does the United States Have a Productivity Slowdown or a Measurement Problem? Brookings Papers on Economic Activity, Spring, 109-157
- Chen Shiyi, 2011, Reconstruction of Sub-industrial Statistical Data in China (1980-2008), China Economic Quarterly (in Chinese), Vol 10(3), 735-776
- Chen Sally, Joong Shik Kang, 2016, Credit Boom, Is China Different? IMF Working Paper 2018/2
- Duarte Margarida and Restuccia Diego, 2017, Relative Prices and Sectoral Productivity, NBER Working Paper No. 23979
- Duernecker Georg, Herrendorf Berthold and Valentinyi Akos, 2017, Structural Change within the Service Sector and the Future of Baumol’s Disease, Manuscript
- Feenstra Robert C., Integration of Trade and Disintegration of Production in the Global Economy, The Journal of Economic Perspectives, 12(4), 31-50
- Griliches Zvi, 1992, Output Measurement in the Service Sectors, National Bureau of Economic Research Studies in Income and Wealth, University of Chicago Press
- Herrendorf Berthold, Rogerson Richard and Valentinyi Ákos, 2014, Growth and Structural Transformation, Handbook of Economic Growth, Volume 2B, ISSN 1574-0684
- International Monetary Fund, 2016, China Article IV Staff Report
- Jain-Chandra Sonali and Zhang Longmei, 2014, How can Korea Boost Potential Output to Ensure Continued Income Convergence? IMF working paper 14/54
- Jorgenson Dale W. and Timmer Marcel P., 2011, Structural Change in Advanced Nations: A New Set of Stylised Facts, The Scandinavian Journal of Economics 113(1), 1–29
- Krugman Paul, 1994, The Myth of Asia's Miracle, Foreign Affairs vol. 73(6), 62-78
- Lam W. Raphael, Schipke Alfred, Tan Yuyan and Tan Zhibo, 2017, Resolving China's Zombies: Tackling Debt and Raising Productivity, IMF Working Paper 17/266
- Lardy Nicholas, 2015, Skeptics of China’s GDP Growth Have Not Made Their Case, Peterson Institute for International Economics: Washington DC, August 14
- Lardy Nicholas, 2016, The Changing Role of the Private Sector in China, Manuscript
- Lin Yifu, Wan Guanghua and Morgan Peter J., 2016, Prospects for a Re-acceleration of Economic Growth in the PRC, Journal of Comparative Economics, 1-12
- Li Haizheng, Liang Yunling, Fraumeni Barbara M., Liu Zhiqiang and Wang Xiaojun, 2013, Human capital in China, 1985-2008, Review of Income and Wealth, 59(2), 212-234
- Lu Yang and Cai Fang, 2014, China’s Shift from the Demographic Dividend to the Reform Dividend, Deepening Reform for China's Long-term Growth and Development book, ANU Press, The Australian National University, Canberra, Australia
- Maliszewski Wojciech and Zhang Longmei, 2015, China’s Growth: Can Goldilocks Outgrow Bears? IMF Working Paper No. 15
- Pritchett Lant and Summers Lawrence H., 2013. Asia-phoria Meet Regression to the Mean, Proceedings, Federal Reserve Bank of San Francisco, issue Nov, 1-35
- Rodrik Dani, 2013. Unconditional Convergence in Manufacturing, The Quarterly Journal of Economics, 165–204
- Rodrik Dani, 2016. Premature Deindustrialization, Journal of Economic Growth, Vol. 21(1), 1-33
- Summers, Lawrence H, 2015, "Demand Side Secular Stagnation." American Economic Review, 105 (5), 60-65
- Syverson, Chad, 2017. Challenges to Mismeasurement Explanations for the US Productivity Slowdown, Journal of Economic Perspectives, 31(2), 165-186
- Triplett Jack E. & Bosworth Barry P., 2003. Productivity Measurement Issues in Services Industries: "Baumol's disease" Has Been Cured, Economic Policy Review, Federal Reserve Bank of New York, Issue Sep., 23-33
- World Bank, 2012, China 2030: Building a Modern, Harmonious and Creative Society. Washington DC: World Bank
- Xu Xianchun, 2004, Problems with China’s GDP accounting in service, Economic Research Journal (in Chinese), volume 3, 20-27
- Yip Winnie & Hsiao William C., 2015, What Drove the Cycles of Chinese Health System Reforms? Health Systems & Reform, 1(1), 52-61
- Young Alwyn (1994), Lessons from the East Asian NICs: A Contrarian View, European Economic Review 38(3), 964-973
- Zhang Longmei, 2016, Rebalancing in China—Progress and Prospects, IMF Working Paper 16/183
- Zhang, L, R. Brooks, D. Ding, H. Ding, H. He, J. Lu and R. Mano, China’s high savings: Drivers, Prospects, and Policies. IMF Working Paper 18/277

### Appendix A: Data — key data construction and adjustments
- Secondary industry data:
  - Data before 2004 is based on Chen (2010). Dataset extended to 2015 using Chen’s method with additional adjustments for consistency, including adjusting for the industrial classification change in 2011, extrapolating from above-scale to all-scale industrial data, and industrial matching for international comparison.
  - Adjustment for industrial classification changes (GB/T 4754 1994 → GB/T4754 2011):
    - Support Activities for Mining: investment mainly driven by Support Activities for Petroleum and Natural Gas (accounts for more than 90 percent). Variable of interest for Support Activities for Mining added back to Extraction of Petroleum and Natural Gas.
    - Manufacture of Rubber and Plastics Products: NBS stopped reporting separately since 2012; reconstructed two series by applying historical ratios. Manufacture of Rubber accounts for about 32 percent of Manufacture of Rubber and Plastics; ratio fairly stable historically.
    - Manufacture of Transport Equipment: taken as sum of Manufacture of Automobiles and Manufacture of Railway, Ship, Aerospace and Other Transport.
    - Repair Service of Metal Products Machinery and Equipment: created after 2011 by consolidating repair service components of multiple manufacturing industries; extrapolated repair services by applying historical shares to the reported consolidated data.
    - Other Industries: defined as sum of Mining of Other Ores, Other Manufacture, and Utilization of Waste Resources. Since 2011, Other Manufacture excludes Manufacture of Artwork (now classified elsewhere); lack of 3-digit data prevents adjustment.
- Adjustment from above-scale to all-scale:
  - After 1998, NBS reports above-scale industrial enterprises only. Adjusted post-1998 above-scale data to all-scale using 2004, 2008 and 2013 census data.
  - Procedure (example for value added):
    - Step 1: Obtain above-scale gross output Q−above and all-scale gross output Q−all in 2004, 2008 and 2013.
    - Step 2: Calculate gross output multiplier Q−all / Q−above for 2004, 2008 and 2013.
    - Step 3: Use linear interpolation to obtain gross output multiplier for other years, producing a time series from 2004 to 2015 at 2-digit level.
    - Same procedure applied to employment data; employment data from China Labor Statistical Yearbook (urban unit employment) scaled to all units using census multipliers.
  - Value-added level estimation (post-2006):
    - Step 1: Obtain growth rate of real value added after 2006, nominal above-scale value added 2004–2015 and Producer Price Indices for Industrial products 2004–2015 from NBS.
    - Step 2: Take 2007 as base year, calculate nominal industrial value added after 2007 using formula 푌푌_t = 푌푌_{t-1} * (P_{t-1} g_t / P_t) in which Y stands for nominal value added, P represents Producer Price Indices, g stands for growth rate of real value added.
    - Step 3: Check consistency by repeating Step 2 for year 2006 and estimating 2007 level; estimated vs actual 2007 level error within 5- 10 percent. Bias may grow over time due to roll-over estimation.
    - Step 4: Deflate estimated nominal value added using Producer Price Indices.
- Adjustment for international comparison:
  - Consolidated 38 NBS industrial sectors to 11 KLEMS sectors (detailed matching listed in Table A1).
- Services data:
  - Source: NBS and KLEMS dataset with adjustments.
  - Nominal value added and employment: aggregated across firm types (urban unit, urban private and self-employed unit, rural private and self-employed, township and village enterprises); shares applied to aggregate service nominal value added and employment reported by NBS under national accounts. NBS classification change in 2003 converted to GB/T4754 2003 for consistency.
  - Price index: NBS publishes price data for five major service sectors; for missing series, substitutes used:
    - Administration and environmental administration: wage index.
    - Scientific research and technology development: average service price index.
    - Others: relevant components of consumer price index.
  - Real value added: computed by dividing revised nominal value added by the price index.
  - For international comparison, converted 14 NBS service sectors to 10 industrial and 11 service sectors based on KLEMS; real value added converted to 2011 international dollars in PPP terms. Other countries’ data from KLEMS.

### Appendix B: Forecast — methodology, baseline and reform scenarios, and aggregation bias
- Forecast approach:
  - Potential GDP growth forecast using supply-side bottom-up approach combined with a convergence framework at the sectoral level.
  - Analyses conducted at a three-sector level and a 21-sector level (10 industrial sectors and 11 service sectors).
- Baseline forecast:
  - Process:
    - Forecast labor productivity for each sector following Bourlès et al. (2010) and Rodrik (2013) convergence specification:
      - ijtjitijtijt DDyy εβ +++=lnˆ
      - ijt yˆ is the growth rate of labor productivity for sector i and country j at time t.
      - ijt y is the level of labor productivity.
      - it D is the industry-year fixed effect and j D is the country fixed effect.
      - ijt ε is the error term.
      - β is the estimated convergence rate.
    - For the service sector, assumption: the same convergence rate as estimated for the industrial by Rodrik (2013), that is 029 .0==industryservice ββ.
    - Forecast growth rate of sectoral labor productivity is given by the equation as specified in the source:
      - )-)ln(lnˆ)ˆ(,,,,frontier it china it frontierf it chinaf it frontierf it chinaf it DDyyyyE（+−+= β
    - Adjusted country fixed effects to reflect historical forecast errors.
    - Aggregate productivity derived by applying forecast sectoral shares (assumed to follow 2012–2016 trend).
    - GDP growth = aggregate labor productivity growth rate + forecast employment growth rate.
  - Applied at both three-sector and detailed KLEMS-based sectoral level.
  - Adjustments made to ensure consistency between sectoral data and macro data; details provided later.
- Reform scenario:
  - To estimate effects of services opening up on productivity, rely on Bourlès et al. (2010) specification where TFP growth depends on initial productivity level and economic reform:
    - ijtjti t ij Frontier ti t ijtij tij Frontier ti Frontier itijt DD y y REGREG y y yy εαααα +++++−+= − − −− − − )ln(*)ln()1(ˆˆ1,1,1,31,3 1,1 01
  - Definitions of ijt yˆ and ijt y same as Rodrik (2013).
  - REG represents sectoral regulation index from OECD product market regulation database.
  - Forecast labor productivity growth rate for the reform scenario follows the source’s specified formula:
    - ) ( )(ln*))(ln1(ˆ )ˆ( ,1 ,1 ,14 ,13 ,1 ,1 0 ,1 ,frontier jt china jt chinaf it frontierf it chinaf it chinaf it chinaf it frontierf it frontierf it chinaf it chinaf it DD y y REGREG y y yyE −+ ++−+= − − −− − − αααα
  - Reform scenario assumptions:
    - Regulation index falls to the lowest level in the OECD sample (lower index means less restrictive regulation).
    - Marginal impact of deregulation based on estimated value in Bourlès et al. (2010).
- Discrepancy between macro and sectoral data and aggregation bias correction:
  - Historical gap: services labor productivity based on macro data vs. 11 subsectors is 14 percent in 2016.
    - Breakdown: 4 percentage points from employment data gap; 10 percentage points from price index used to deflate nominal value added.
  - Industrial sector bias larger because sectoral value added and employment do not sum to the aggregate national account figures. Industrial employment adjustment from above-scale to all-scale may remain insufficient.
  - Correction method:
    - Compute average aggregation bias in 2011–2015 as average difference between historical data and in-sample forecast of labor productivity growth from 2011–2015.
    - Assume aggregation bias for both industry and services is largest in 2016 and declines linearly to zero by end of forecast period in 2030.
    - Add aggregation bias to the model forecast to obtain final GDP growth rate.

*Italic: Source: wpiea2019263-print-pdf - References (IMF PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019263-print-pdf.pdf_