## wpiea2019117 - Section 4 explores the implications of structural changes in the industrial structure for economic

## Source details

**Canonical URL:** [wpiea2019117 - Section 4 explores the implications of structural changes in the industrial structure for economic](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019117.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019117.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019117.pdf.json)

---

### Related Literature
- Focus areas:
  - (i) propagation of microeconomic shocks to the economy through input-output network linkages;
  - (ii) business cycle (BC) comovement through trade linkages;
  - (iii) economic diversification (or specialization) and development.
- Domestic propagation: shocks to firms or sectors can spread through input-output linkages producing larger macroeconomic impacts (examples cited: Acemoglu et al. 2012, 2016; Carvalho and Gabaix 2013; Foerster et al. 2011).
- External propagation: trade linkages transmit external shocks across borders (example cited: Acemoglu et al. 2016).
- Comparison to Acemoglu et al. (2016):
  - Acemoglu et al. (2016) examined both direct (first-order) and indirect (higher-order) vertical linkages for the US.
  - This paper considers only direct upstream and downstream linkages across industries, as in di Giovanni and Levchenko (2010) and di Giovanni et al. (2018).
  - Rationale: indirect linkages can matter but direct vertical linkages are likely to comprise a dominant channel in most cases.
- International BC comovement:
  - Frankel and Rose (1998) documented that countries that trade more exhibit higher BC correlation.
  - Shea (2002) found input-output linkages, not common shocks, were important for sectoral comovement in the U.S.
  - Industry- and firm-level studies (di Giovanni and Levchenko 2010; di Giovanni et al. 2018) show positive international trade–comovement relationships driven by transmission of shocks through trade and vertical linkages.
  - Lee (2019b) found export linkages were important for Korea’s international BC comovement and that increased trade with China contributed most to aggregate BC comovement.
- Economic diversification literature:
  - Two strands: (a) development → diversification (hump- or U-shaped pattern: diversify then respecialize) (examples: Imbs and Wacziarg 2003; Koren and Tenreyro 2007; Cadot et al. 2011); (b) diversification → growth, with export diversification stabilizing export earnings and domestic outputs (examples: Jansen 2004; Cavallo et al. 2008).
  - Surveys: Cadot et al. (2013) and Mau (2016).
  - Cadot et al. (2011) noted extensive margin drives the U-shaped pattern.
- Paper’s contribution:
  - Modifies empirical model of Acemoglu et al. (2016) to explore transmission of domestic and external shocks through vertical and trade linkages in Korea.
  - Korea is noted as highly dependent on international trade and highly interconnected in upstream and downstream markets.
  - Analysis uses Korean industry-level data to shed light on transmission mechanisms of domestic and external shocks.

### Data and Descriptive Statistics
- Data sources and coverage:
  - Korea’s national accounts: 1970–2017.
  - WIOD (2016 release): 2000–2014; contains World Input-Output Tables (WIOT) and Socio-Economic Accounts (SEA).
  - WIOT covers 56 industries for each of the 43 countries over 15 years (2000–2014).
- Empirical focus:
  - Transmission channels: (i) idiosyncratic industry shocks (domestic); (ii) foreign country shocks (external).
  - Foreign shocks focused on Korea’s three largest trading partners: China, the U.S., and Japan.
  - Industry aggregation: combined or dropped some industries for reliable data, resulting in a total of 38 industries, including 14 manufacturing industries.
  - Econometric analysis uses industry-level data on input, output, value added, and international trade from WIOD.
- Notes on data treatment:
  - Some industries with identical growth rates of industry value added were combined; suggestion that original separate value-added levels were unavailable and extrapolation was used.

### Industrial Structure: Concentration and Interconnectedness
- Sectoral composition and trends:
  - Manufacturing and services account for most of Korea’s gross domestic product (GDP).
  - Manufacturing share in GDP rose to 31.6 percent, on average, in the 2010s from 12.7 percent in the 1970s.
  - Services share in GDP has remained stable at around 60 percent.
  - Other sectors (e.g., non-manufacturing and agriculture) have declined over time.
  - Combined share of manufacturing and services reached over 90 percent of GDP in 2017.
- Concentration within sectors:
  - Manufacturing is highly concentrated in a few key industries; services are less concentrated.
  - Manufacturing exhibits a U-shaped relationship between economic development and diversification:
    - Share of the largest three industries in manufacturing declined to 41.9 percent in the 1990s, but rose to 57.6 percent in the 2010s.
  - In services, the share of the top three industries declined over recent decades, reaching 39.5 percent in the 2010s.
  - Largest three manufacturing industries over time:
    - 1970s and 1980s: Food, textiles, chemicals, and basic metals were the largest three.
    - Since the 2000s: Electronics, transportation, and chemicals have been the main manufacturing industries.
  - Largest three service industries over time: Wholesale and retail trade, real estate, and public administration and defense have been the largest three industries in services over the last several decades.
- Vertical linkages (domestic interconnectedness):
  - Domestic vertical linkages (share of intermediate input in total output) increased to 47.1 percent in 2009 from 44.3 percent in 2000, then remained stable afterwards.
  - Upstream linkage definition: share of intermediate input supply (purchase) in sectoral or industrial gross output; labels adopted in this paper:
    - Upstream linkage: connectedness to buyers of an industry (shocks to a buyer flow up the input-output network).
    - Downstream linkage: connectedness to sellers of an industry (shocks to a seller flow down the input-output network).
  - Sector-level trends:
    - Upstream and downstream linkages in manufacturing increased until the global financial crisis (GFC) and then started to decline.
    - Vertical linkages remained stable in services.
  - Industry-level trends for top three manufacturing industries since the 2000s (electronics, transportation, chemicals):
    - Upstream and downstream linkages declined slightly in the aftermath of the GFC.
    - Upstream linkages are highest in chemicals and almost twice higher than in electronics and transportation.
    - Downstream linkages are similar across the three industries at around 60 percent.
  - Notes on industry classification:
    - Electronics includes “Computer, electronic and optical products” and “Electrical equipment”.
    - Chemicals includes “Chemical and chemical products” and “Basic pharmaceutical products and preparations”.
    - Transportation includes “Motor vehicles, trailers and semi-trailers” and “Other transport equipment”.
- Trade linkages (global interconnectedness):
  - Export and import linkage measurement: share of gross export (imports) to (from) other countries in sectoral or industrial gross output.
  - Post-GFC change: share of exports and imports in total output increased by about 5 percentage points each in the aftermath of the GFC.
  - Sector-level patterns:
    - Export and import linkages substantially higher in manufacturing than other sectors.
    - Sharp rise of export and import linkages in manufacturing since the 2000s.
    - In services, export and import linkages remained stable at around 5 percent and 3 percent, respectively.
  - Industry-level patterns:
    - Export and import linkages rose in major manufacturing industries after the GFC.
    - Electronics and transportation show export linkages about two times higher than in chemicals.
    - Import linkages are highest in electronics.
  - Bilateral trade partner patterns:
    - Export and import linkages to China rose drastically in major manufacturing industries throughout the 2000s.
    - Trade linkages to the U.S. and Japan declined noticeably in major manufacturing industries over the same period.
    - Electronics experienced the most drastic structural changes in trade linkages: export and import linkages to China rose significantly while trade linkages to the U.S. declined substantially over the last 15 years.
    - Export and import exposures to China, the U.S., and Japan account for almost 50 percent of total trade linkages in the major manufacturing industries, except for export linkages in the transportation industry.
  - Contextual note: The increase in trade linkages to China in the 2000s may relate to China’s growing presence in the global economy, especially after China joined the World Trade Organization (WTO) in 2001.

### Macroeconomic Development: Economic Growth and Volatility
- Sectoral contributions to GDP growth:
  - Manufacturing and services have been key contributors to economic growth.
  - Contribution of manufacturing and services to GDP growth has been declining, but these sectors still account for most GDP growth.
  - Decline in contribution to growth observed across all industry sectors.
- Volatility trends:
  - Volatility of value-added growth remained high until the 1990s across sectors, then declined afterwards.
  - The high volatility of the 1990s partly reflects large swings during the Asian Financial Crisis in the late 1990s.
  - Standard deviation (volatility) of growth declined substantially through the 2000s, consistent with overall GDP volatility trends.
  - Volatility level has been higher in manufacturing than in services and other sectors through the 2000s, potentially related to higher concentration and interconnectedness in manufacturing.
- Industrial decomposition of growth:
  - Electronics contributed almost half of manufacturing growth since the 2000s.
  - Growth contribution of the largest three manufacturing industries:
    - 1980s: 26.5 percent.
    - 2000s: almost 70 percent.
  - Service sector growth contribution is less concentrated; share of the major three industries’ contribution has remained around 30 to 40 percent since the 1980s.
- Industry-level volatility:
  - Electronics and transportation (the two largest share manufacturing industries through the 2000s) have been relatively volatile compared to other manufacturing and service industries.
  - Implication: potential factors of economic instability may still be embedded in the Korean economy despite recent declines in overall growth volatility.

### Empirical specification and data (Transmission of Domestic and External Shocks)
- Focus: role of domestic vertical linkages and direct/indirect trade linkages in transmitting shocks to industry labor productivity growth.
- Data: industry-level international input-output data of the WIOD (2000–2014).
- Domestic linkage measures (Equations (1)–(2)):
  - UP_i,t^DM = ∑ IO_i,j,t × ∆ln Y_j,t
  - DN_i,t^DM = ∑ IO_j,i,t × ∆ln Y_j,t
- External linkage measures (Equations (3)–(6)):
  - OWN_i,t^EX = ∑ EX_i,C,t × ∆ln Y_C,t
  - OWN_i,t^IM = ∑ IM_i,C,t × ∆ln Y_C,t
  - UP_i,t^EX = ∑∑ IO_i,j,t × EX_j,C,t_jC × ∆ln Y_C,t
  - DN_i,t^IM = ∑∑ IO_j,i,t × IM_i,C,t_jC × ∆ln Y_C,t
- Estimating equations for labor productivity growth (∆ln LP_it) include:
  - lagged ln(LP), ∆ln K_it (capital/labor growth), industry fixed effects (μ_i), sector-year or year fixed effects (δ_st), and error term ε_it (Equations (7)–(9)).

### Estimation results — main findings
- General:
  - Domestic industry shocks propagate more strongly via downstream linkages (seller/supplier shocks) than via upstream linkages (buyer/customer shocks).
  - External country shocks are propagated mainly through Korea’s own (direct) export linkages.
  - Coefficient signs and magnitudes are largely consistent across specifications.
- Controls:
  - Lagged log(labor productivity): negative and significant in most specifications.
  - Capital/labor growth: positive and significant in all specifications.

### Key estimated coefficients (selected, specification [3] — both domestic and external shocks)
- Total industry (specification [3]):
  - Lagged log(labor productivity): -5.997**
  - Capital/labor growth: 0.556***
  - Upstream domestic shock: 0.767
  - Downstream domestic shock: 0.795***
  - Own export shock: 4.490
- Manufacturing (specification [3]):
  - Lagged log(labor productivity): -4.146
  - Capital/labor growth: 0.659***
  - Upstream domestic shock: 1.031**
  - Downstream domestic shock: 0.734**
  - Own export shock: 5.843*
- Significance markers: ***, **, * indicate levels of significance at 1%, 5%, 10%, respectively.

### Quantitative estimates of productivity impacts from one standard deviation shocks
- Method: use significant coefficients from specification [3] and industry-average upstream/downstream/export linkages to compute average impacts of shocks (Equations (10)–(11)); shock sizes set to one standard deviation of industry value added growth (top three manufacturing industries: electronics, transportation, chemicals) and one standard deviation of country GDP growth (China, the U.S., Japan) for 2001−2014.
- Findings (average impact on productivity growth):
  - Domestic industry shocks (total industry):
    - Electronics: impact (average) = 0.21 (percentage point)
    - Transportation: impact (average) = 0.14 (percentage point)
    - Chemicals: impact (average) = 0.52 (percentage point)
  - External country shocks (total industry):
    - China: impact = 0.26 (percentage point)
    - U.S.: impact = 0.12 (percentage point)
    - Japan: impact = 0.11 (percentage point)
  - Manufacturing sector impacts (about two times larger than total industry impacts):
    - Electronics: impact = 0.62 (percentage point)
    - Transportation: impact = 0.54 (percentage point)
    - Chemicals: impact = 0.77 (percentage point)
    - China: impact = 0.70 (percentage point)
    - U.S.: impact = 0.29 (percentage point)
    - Japan: impact = 0.27 (percentage point)
- Aggregate/overall impacts (sum of estimated impacts from individual shocks):
  - Overall impact of each one standard deviation of domestic and external shocks:
    - Total industry: 1.4 percentage points (on average)
    - Manufacturing: 3.2 percentage points (on average)

### Interpretation and implications
- Structural drivers of amplification:
  - High concentration and interconnectedness—dominated by a few volatile, large firms and heavy international trade—amplify transmission of shocks through vertical and trade linkages.
- Industry-specific risks:
  - Chemicals emerges as the most volatile among the three key manufacturing industries and thus the largest domestic source of productivity impact (0.52 percentage point for total industry).
- External exposure:
  - China’s growth shocks pose the largest external threat to Korea’s productivity among the three trading partners (0.26 percentage point for total industry), reflecting strong trade linkages.
- Sectoral asymmetry:
  - Manufacturing’s stronger vertical and trade linkages explain why shocks produce roughly double the productivity impact in manufacturing compared with the total industry.
- Policy-relevant implication:
  - High trade linkages to China and the U.S. increase Korea’s vulnerability to unfavorable economic episodes in those countries; trade tensions affecting China and the U.S. could have sizable direct and indirect effects on Korea through export channels and subsequent upstream/downstream propagation across industries.

### Robustness checks (selected)
- Alternative specifications:
  - Transmission via top five manufacturing industries: downstream linkages remain the main propagation channel; direct export propagation estimated as insignificant but broadly consistent in sign and magnitude with baseline.
  - External country shocks considered separately for China, the U.S., and Japan: confirm important role of downstream linkages and direct export linkages for China and Japan; for the U.S., domestic downstream effects are significant while direct export effects are negative but insignificant for the total industry.
  - Alternative export linkage measure using intermediate input exports: results broadly align with baseline—domestic shocks have larger downstream effects for total industry; for manufacturing both upstream and downstream linkages significant; external shocks propagate mainly through direct export linkages.
- Additional specific robustness estimates (total industry specification [3] with country-specific shocks):
  - Own export shock (China): 7.072***
  - Own export shock (U.S.): -8.470 (not significant)
  - Own export shock (Japan): 49.09** (large point estimate with large standard error)
  - Upstream domestic shock (manufacturing, country-specific): for U.S. and Japan coefficients become significant in some specifications (e.g., 0.865*, 0.964** in manufacturing for U.S. and Japan respectively in some columns).

### Concluding Remarks and Policy-Relevant Points
- Overview:
  - Throughout the 2000s, the Korean economy became more concentrated in a few manufacturing industries, while its interconnectedness across industries and to foreign countries rose via vertical relationships and trade linkages.
  - Dominant industries are highly interconnected with other domestic industries via upstream/downstream linkages and with foreign markets via export/import linkages, and are dominated by a few large firms.
  - The rise of economic concentration and interconnectedness could become sources of macroeconomic instability.
- Key empirical conclusions:
  - Domestic industry shocks have larger downstream effects than upstream effects, implying industries are more likely to be affected by the seller’s growth shocks than by the buyer’s growth shocks.
  - External country shocks are propagated to Korean industries mainly through direct export linkages.
  - Growth shocks in key manufacturing industries and/or in major trading partners can lead to large swings in the overall economy due to transmission through vertical and trade linkages.
- Quantitative summary:
  - Estimated overall productivity impacts of each one standard deviation of growth shocks (top three manufacturing industries and major three trading partners): 1.4 percentage points, on average, for the total industry; 3.2 percentage points for manufacturing.
  - Growth shocks in chemicals have the largest productivity impacts among domestic industries.
  - External country shocks: China’s growth shocks have the largest productivity impacts.
- Regression highlights (selected estimates reproduced):
  - Lagged log(labor productivity): -6.156** (External), -5.757** (Both), -3.708 (External, smaller sample), -3.802 (Both, smaller sample). (Standard errors: (2.459), (2.579), (2.511), (2.693))
  - Capital/labor growth: 0.572***, 0.561***, 0.714***, 0.665***. (Standard errors: (0.056), (0.058), (0.107), (0.107))
  - Upstream domestic shock: 0.116 (External), 0.954** (Both). (Standard errors: (0.429), (0.398))
  - Downstream domestic shock: 1.672*** (External), 0.832** (Both). (Standard errors: (0.366), (0.281))
  - Own export shock: 5.246, 7.755*, 4.916, 7.723. (Standard errors: (4.376), (3.923), (5.218), (4.529))
  - Own import shock: -1.026, 4.349, 2.581, 15.16. (Standard errors: (16.39), (16.09), (20.07), (19.38))
  - Upstream export shock: -0.021, -0.020, -0.008, -0.046. (Standard errors: (0.046), (0.052), (0.079), (0.062))
  - Downstream import shock: 0.470, 0.225, 0.296, -0.254. (Standard errors: (0.519), (0.447), (0.728), (0.675))
  - Observations: 532 (full sample specifications), 196 (smaller sample specifications).
  - R2-within: 0.663, 0.694, 0.690, 0.726.
  - Number of industries: 38 (full sample), 14 (manufacturing sample).
- Extensions and further research:
  - Extend analysis to value-added exports versus gross exports to better assess transmission of external country growth shocks.
  - Consider higher-order interconnectedness across industries to capture potential “cascade effects” whereby shocks propagate beyond immediate downstream customers.

*Source: wpiea2019117 - Section 4 explores the implications of structural changes in the industrial structure for economic (IMF Working Paper).*

### Section 4 explores the implications of structural changes in the industrial structure for economic

### wpiea2019117 - Section 4 explores the implications of structural changes in the industrial structure for economic

### Related Literature
- Focus areas: (i) propagation of microeconomic shocks to the economy through input-output network linkages; (ii) business cycle (BC) comovement through trade linkages; and (iii) economic diversification (or specialization) and development.
- Domestic propagation literature: shocks to firms or sectors can spread through input-output linkages producing larger macroeconomic impacts (examples cited: Acemoglu et al. 2012, 2016; Carvalho and Gabaix 2013; Foerster et al. 2011).
- External propagation literature: trade linkages transmit external shocks across borders (example cited: Acemoglu et al. 2016).
- Comparison to Acemoglu et al. (2016):
  - Acemoglu et al. (2016) examined both direct (first-order) and indirect (higher-order) vertical linkages for the US.
  - This paper considers only direct upstream and downstream linkages across industries, as in di Giovanni and Levchenko (2010) and di Giovanni et al. (2018).
  - Rationale: indirect linkages can matter but direct vertical linkages are likely to comprise a dominant channel in most cases.
- International BC comovement literature:
  - Frankel and Rose (1998) documented that countries that trade more exhibit higher BC correlation.
  - Shea (2002) found input-output linkages, not common shocks, were important for sectoral comovement in the U.S.
  - Industry- and firm-level studies (di Giovanni and Levchenko 2010; di Giovanni et al. 2018) show positive international trade–comovement relationships driven by transmission of shocks through trade and vertical linkages.
  - Complementary study: Lee (2019b) found export linkages were important for Korea’s international BC comovement and that increased trade with China contributed most to aggregate BC comovement.
- Economic diversification literature:
  - Two strands: (a) development → diversification (hump- or U-shaped pattern: diversify then respecialize) (examples: Imbs and Wacziarg 2003; Koren and Tenreyro 2007; Cadot et al. 2011); (b) diversification → growth, with export diversification stabilizing export earnings and domestic outputs (examples: Jansen 2004; Cavallo et al. 2008).
  - Surveys: Cadot et al. (2013) and Mau (2016).
  - Cadot et al. (2011) noted extensive margin drives the U-shaped pattern.
- Paper’s contribution:
  - Modifies empirical model of Acemoglu et al. (2016) to explore transmission of domestic and external shocks through vertical and trade linkages in Korea.
  - Korea is noted as highly dependent on international trade and highly interconnected in upstream and downstream markets.
  - Analysis uses Korean industry-level data to shed light on transmission mechanisms of domestic and external shocks.

### Data and Descriptive Statistics
- Data sources and coverage:
  - Korea’s national accounts: 1970–2017.
  - WIOD (2016 release): 2000–2014; contains World Input-Output Tables (WIOT) and Socio-Economic Accounts (SEA).
  - WIOT covers 56 industries for each of the 43 countries over 15 years (2000–2014).
- Empirical focus:
  - Transmission channels: (i) idiosyncratic industry shocks (domestic); (ii) foreign country shocks (external).
  - Foreign shocks focused on Korea’s three largest trading partners: China, the U.S., and Japan.
  - Industry aggregation: combined or dropped some industries for reliable data, resulting in a total of 38 industries, including 14 manufacturing industries.
  - Econometric analysis uses industry-level data on input, output, value added, and international trade from WIOD.
- Notes on data treatment:
  - Some industries with identical growth rates of industry value added were combined; suggestion that original separate value-added levels were unavailable and extrapolation was used.

### Industrial Structure: Concentration and Interconnectedness
- Sectoral composition and trends:
  - Manufacturing and services account for most of Korea’s gross domestic product (GDP).
  - Manufacturing share in GDP rose to 31.6 percent, on average, in the 2010s from 12.7 percent in the 1970s.
  - Services share in GDP has remained stable at around 60 percent.
  - Other sectors (e.g., non-manufacturing and agriculture) have declined over time.
  - Combined share of manufacturing and services reached over 90 percent of GDP in 2017.
- Concentration within sectors:
  - Manufacturing is highly concentrated in a few key industries; services are less concentrated.
  - Manufacturing exhibits a U-shaped relationship between economic development and diversification:
    - Share of the largest three industries in manufacturing declined to 41.9 percent in the 1990s, but rose to 57.6 percent in the 2010s.
  - In services, the share of the top three industries declined over recent decades, reaching 39.5 percent in the 2010s.
  - Largest three manufacturing industries over time:
    - 1970s and 1980s: Food, textiles, chemicals, and basic metals were the largest three.
    - Since the 2000s: Electronics, transportation, and chemicals have been the main manufacturing industries.
  - Largest three service industries over time: Wholesale and retail trade, real estate, and public administration and defense have been the largest three industries in services over the last several decades.
- Vertical linkages (domestic interconnectedness):
  - Domestic vertical linkages (share of intermediate input in total output) increased to 47.1 percent in 2009 from 44.3 percent in 2000, then remained stable afterwards.
  - Upstream linkage definition: share of intermediate input supply (purchase) in sectoral or industrial gross output; labels adopted in this paper:
    - Upstream linkage: connectedness to buyers of an industry (shocks to a buyer flow up the input-output network).
    - Downstream linkage: connectedness to sellers of an industry (shocks to a seller flow down the input-output network).
  - Sector-level trends:
    - Upstream and downstream linkages in manufacturing increased until the global financial crisis (GFC) and then started to decline.
    - Vertical linkages remained stable in services.
  - Industry-level trends for top three manufacturing industries since the 2000s (electronics, transportation, chemicals):
    - Upstream and downstream linkages declined slightly in the aftermath of the GFC.
    - Upstream linkages are highest in chemicals and almost twice higher than in electronics and transportation.
    - Downstream linkages are similar across the three industries at around 60 percent.
  - Notes on industry classification:
    - Electronics includes “Computer, electronic and optical products” and “Electrical equipment”.
    - Chemicals includes “Chemical and chemical products” and “Basic pharmaceutical products and preparations”.
    - Transportation includes “Motor vehicles, trailers and semi-trailers” and “Other transport equipment”.
- Trade linkages (global interconnectedness):
  - Export and import linkage measurement: share of gross export (imports) to (from) other countries in sectoral or industrial gross output.
  - Post-GFC change: share of exports and imports in total output increased by about 5 percentage points each in the aftermath of the GFC.
  - Sector-level patterns:
    - Export and import linkages substantially higher in manufacturing than other sectors.
    - Sharp rise of export and import linkages in manufacturing since the 2000s.
    - In services, export and import linkages remained stable at around 5 percent and 3 percent, respectively.
  - Industry-level patterns:
    - Export and import linkages rose in major manufacturing industries after the GFC.
    - Electronics and transportation show export linkages about two times higher than in chemicals.
    - Import linkages are highest in electronics.
  - Bilateral trade partner patterns:
    - Export and import linkages to China rose drastically in major manufacturing industries throughout the 2000s.
    - Trade linkages to the U.S. and Japan declined noticeably in major manufacturing industries over the same period.
    - Electronics experienced the most drastic structural changes in trade linkages: export and import linkages to China rose significantly while trade linkages to the U.S. declined substantially over the last 15 years.
    - Export and import exposures to China, the U.S., and Japan account for almost 50 percent of total trade linkages in the major manufacturing industries, except for export linkages in the transportation industry.
  - Contextual note: The increase in trade linkages to China in the 2000s may relate to China’s growing presence in the global economy, especially after China joined the World Trade Organization (WTO) in 2001.

### Macroeconomic Development: Economic Growth and Volatility
- Sectoral contributions to GDP growth:
  - Manufacturing and services have been key contributors to economic growth.
  - Contribution of manufacturing and services to GDP growth has been declining, but these sectors still account for most GDP growth.
  - Decline in contribution to growth observed across all industry sectors.
- Volatility trends:
  - Volatility of value-added growth remained high until the 1990s across sectors, then declined afterwards.
  - The high volatility of the 1990s partly reflects large swings during the Asian Financial Crisis in the late 1990s.
  - Standard deviation (volatility) of growth declined substantially through the 2000s, consistent with overall GDP volatility trends.
  - Volatility level has been higher in manufacturing than in services and other sectors through the 2000s, potentially related to higher concentration and interconnectedness in manufacturing.
- Industrial decomposition of growth:
  - Electronics contributed almost half of manufacturing growth since the 2000s.
  - Growth contribution of the largest three manufacturing industries:
    - 1980s: 26.5 percent.
    - 2000s: almost 70 percent.
  - Service sector growth contribution is less concentrated; share of the major three industries’ contribution has remained around 30 to 40 percent since the 1980s.
- Industry-level volatility:
  - Electronics and transportation (the two largest share manufacturing industries through the 2000s) have been relatively volatile compared to other manufacturing and service industries.
  - Implication: potential factors of economic instability may still be embedded in the Korean economy despite recent declines in overall growth volatility.

*Source: wpiea2019117 - Section 4 explores the implications of structural changes in the industrial structure for economic (IMF PDF content provided).*

### 4. Transmission of Domestic and External Shocks

### 4. Transmission of Domestic and External Shocks

### Empirical specification and data
- Focus: role of domestic vertical linkages and direct/indirect trade linkages in transmitting shocks to industry labor productivity growth.
- Data: industry-level international input-output data of the WIOD (2000–2014).
- Domestic linkage measures (Equations (1)–(2)):
  - UP_i,t^DM = ∑ IO_i,j,t × ∆ln Y_j,t
  - DN_i,t^DM = ∑ IO_j,i,t × ∆ln Y_j,t
- External linkage measures (Equations (3)–(6)):
  - OWN_i,t^EX = ∑ EX_i,C,t × ∆ln Y_C,t
  - OWN_i,t^IM = ∑ IM_i,C,t × ∆ln Y_C,t
  - UP_i,t^EX = ∑∑ IO_i,j,t × EX_j,C,t_jC × ∆ln Y_C,t
  - DN_i,t^IM = ∑∑ IO_j,i,t × IM_i,C,t_jC × ∆ln Y_C,t
- Estimating equations for labor productivity growth (∆ln LP_it) include lagged ln(LP), ∆ln K_it (capital/labor growth), industry fixed effects (μ_i), sector-year or year fixed effects (δ_st), and error term ε_it (Equations (7)–(9)).

### Estimation results — main findings
- General:
  - Domestic industry shocks propagate more strongly via downstream linkages (seller/supplier shocks) than via upstream linkages (buyer/customer shocks).
  - External country shocks are propagated mainly through Korea’s own (direct) export linkages.
  - Coefficient signs and magnitudes are largely consistent across specifications.
- Controls behave as expected:
  - Lagged log(labor productivity): negative and significant in most specifications.
  - Capital/labor growth: positive and significant in all specifications.

### Key estimated coefficients (selected, specification [3] — both domestic and external shocks)
- Total industry (specification [3]):
  - Lagged log(labor productivity): -5.997**
  - Capital/labor growth: 0.556***
  - Upstream domestic shock: 0.767
  - Downstream domestic shock: 0.795***
  - Own export shock: 4.490
- Manufacturing (specification [3]):
  - Lagged log(labor productivity): -4.146
  - Capital/labor growth: 0.659***
  - Upstream domestic shock: 1.031**
  - Downstream domestic shock: 0.734**
  - Own export shock: 5.843*
- Notes on significance markers in the tables: ***, **, * indicate levels of significance at 1%, 5%, 10%, respectively.

### Quantitative estimates of productivity impacts from one standard deviation shocks
- Method: use significant coefficients from specification [3] and industry-average upstream/downstream/export linkages to compute average impacts of shocks (Equations (10)–(11)); shock sizes set to one standard deviation of industry value added growth (top three manufacturing industries: electronics, transportation, chemicals) and one standard deviation of country GDP growth (China, the U.S., Japan) for 2001−2014.
- Findings (average impact on productivity growth):
  - Domestic industry shocks (total industry):
    - Electronics: impact (average) = 0.21 (percentage point)
    - Transportation: impact (average) = 0.14 (percentage point)
    - Chemicals: impact (average) = 0.52 (percentage point)
  - External country shocks (total industry):
    - China: impact = 0.26 (percentage point)
    - U.S.: impact = 0.12 (percentage point)
    - Japan: impact = 0.11 (percentage point)
  - Manufacturing sector impacts are about two times larger than total industry impacts:
    - Manufacturing (selected impacts shown in table):
      - Electronics: impact = 0.62 (percentage point)
      - Transportation: impact = 0.54 (percentage point)
      - Chemicals: impact = 0.77 (percentage point)
      - China: impact = 0.70 (percentage point)
      - U.S.: impact = 0.29 (percentage point)
      - Japan: impact = 0.27 (percentage point)
- Aggregate/overall impacts (sum of estimated impacts from individual shocks):
  - Overall impact of each one standard deviation of domestic and external shocks:
    - Total industry: 1.4 percentage points (on average)
    - Manufacturing: 3.2 percentage points (on average)

### Interpretation and implications
- The high concentration and interconnectedness of Korea’s industrial structure—dominated by a few volatile, large firms and heavily dependent on international trade—amplify transmission of shocks through vertical and trade linkages.
- Chemicals emerges as the most volatile among the three key manufacturing industries and thus the largest domestic source of productivity impact (0.52 percentage point for total industry).
- China’s growth shocks pose the largest external threat to Korea’s productivity among the three trading partners (0.26 percentage point for total industry), reflecting strong trade linkages.
- Manufacturing’s stronger vertical and trade linkages explain why shocks produce roughly double the productivity impact in manufacturing compared with the total industry.
- Policy-relevant implication: high trade linkages to China and the U.S. increase Korea’s vulnerability to unfavorable economic episodes in those countries; trade tensions affecting China and the U.S. could have sizable direct and indirect effects on Korea through export channels and subsequent upstream/downstream propagation across industries.

### Robustness checks
- Alternative specifications examined:
  - Transmission via top five manufacturing industries: downstream linkages remain the main propagation channel; direct export propagation estimated as insignificant but broadly consistent in sign and magnitude with baseline.
  - External country shocks considered separately for China, the U.S., and Japan (Table 4): results confirm the important role of downstream linkages and direct export linkages for China and Japan; for the U.S., domestic downstream effects are significant while direct export effects are negative but insignificant for the total industry.
  - Alternative export linkage measure using intermediate input exports (to mitigate double-counting from gross exports): results broadly align with baseline—domestic shocks have larger downstream effects for total industry; for manufacturing both upstream and downstream linkages significant; external shocks propagate mainly through direct export linkages.
- Additional specific robustness estimates (selected, total industry specification [3] with country-specific shocks):
  - Own export shock (China): 7.072***
  - Own export shock (U.S.): -8.470 (not significant)
  - Own export shock (Japan): 49.09** (note: large point estimate with large standard error)
  - Upstream domestic shock (manufacturing, country-specific): for U.S. and Japan coefficients become significant in some specifications (e.g., 0.865*, 0.964** in manufacturing for U.S. and Japan respectively in some columns).

*Source: wpiea2019117 - 4. Transmission of Domestic and External Shocks (IMF Working Paper).*

### 5. Concluding Remarks

### 5. Concluding Remarks

### Overview
- Throughout the 2000s, the Korean economy became more concentrated in a few manufacturing industries, while its interconnectedness across industries and to foreign countries rose via vertical relationships and trade linkages.
- Dominant industries are highly interconnected with other domestic industries via upstream/downstream linkages and with foreign markets via export/import linkages, and are dominated by a few large firms.
- The rise of economic concentration and interconnectedness could become sources of macroeconomic instability.

### Empirical approach
- The paper uses industry-level international input-output data and an extended model of Acemoglu et al. (2016) adapted to the Korean economy.
- The analysis investigates the role of vertical and trade linkages in transmitting growth shocks from domestic and external sources.

### Key findings
- Domestic industry shocks have larger downstream effects than upstream effects, implying industries are more likely to be affected by the seller’s growth shocks than by the buyer’s growth shocks.
- External country shocks are propagated to Korean industries mainly through direct export linkages, implying industries highly involved in exports made directly to foreign countries are significantly affected by the country’s growth shocks.
- Growth shocks in key manufacturing industries and/or in major trading partners can lead to large swings in the overall economy due to transmission through vertical and trade linkages.

### Quantitative impacts and regression highlights
- Estimated overall productivity impacts of each one standard deviation of growth shocks:
  - Top three manufacturing industries (electronics, transportation, chemicals) and major three trading partners (China, the U.S., Japan): 1.4 percentage points, on average, for the total industry.
  - For manufacturing: 3.2 percentage points.
- Growth shocks in chemicals have the largest productivity impacts among domestic industries, mainly due to the industry’s high vertical linkages and high volatility.
- External country shocks: China’s growth shocks have the largest productivity impacts owing to Korea’s strong trade linkage with China.

- Selected regression coefficient estimates (Dependent variable: Labor productivity growth):
  - Lagged log(labor productivity): -6.156** (External), -5.757** (Both), -3.708 (External, smaller sample), -3.802 (Both, smaller sample). (Standard errors: (2.459), (2.579), (2.511), (2.693))
  - Capital/labor growth: 0.572***, 0.561***, 0.714***, 0.665***. (Standard errors: (0.056), (0.058), (0.107), (0.107))
  - Upstream domestic shock: 0.116 (External), 0.954** (Both). (Standard errors: (0.429), (0.398))
  - Downstream domestic shock: 1.672*** (External), 0.832** (Both). (Standard errors: (0.366), (0.281))
  - Own export shock: 5.246, 7.755*, 4.916, 7.723. (Standard errors: (4.376), (3.923), (5.218), (4.529))
  - Own import shock: -1.026, 4.349, 2.581, 15.16. (Standard errors: (16.39), (16.09), (20.07), (19.38))
  - Upstream export shock: -0.021, -0.020, -0.008, -0.046. (Standard errors: (0.046), (0.052), (0.079), (0.062))
  - Downstream import shock: 0.470, 0.225, 0.296, -0.254. (Standard errors: (0.519), (0.447), (0.728), (0.675))
- Regression design and fit notes:
  - Industry fixed effect: Yes in all specifications.
  - Sector-Year fixed effect: Yes in two specifications, No in two (smaller sample).
  - Year fixed effect: No in two specifications, Yes in two (smaller sample).
  - Observations: 532 (full sample specifications), 196 (smaller sample specifications).
  - R2-within: 0.663, 0.694, 0.690, 0.726.
  - Number of industries: 38 (full sample), 14 (manufacturing sample).
  - Notes: Constant is included in all specifications. ***, **, * indicate levels of significance at 1%, 5%, 10%, respectively.

### Extensions and further research
- Extend analysis to an alternative measure of trade linkages using value-added exports, which may be more relevant for assessing transmission of external country growth shocks.
- Consider differences between gross exports (used in this paper) and value-added exports documented in the literature (e.g., Foster-McGregor and Stehrer 2013; Johnson 2014; Koopman et al. 2014).
- Extend to higher-order interconnectedness across industries to capture potential “cascade effects” whereby shocks propagate beyond immediate downstream customers (e.g., Acemoglu et al. 2012, 2016).

*Source: wpiea2019117 - 5. Concluding Remarks*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019117.pdf_